<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Tanya Thakur on Medium]]></title>
        <description><![CDATA[Stories by Tanya Thakur on Medium]]></description>
        <link>https://medium.com/@tanyathakur6?source=rss-e221ca4e11c4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*oCOc6sWC7WMQXN1XEIhArA@2x.jpeg</url>
            <title>Stories by Tanya Thakur on Medium</title>
            <link>https://medium.com/@tanyathakur6?source=rss-e221ca4e11c4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 09:40:06 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@tanyathakur6/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Demo on Building a Chatbot with Alter NLU]]></title>
            <link>https://medium.com/kontikilabs/demo-on-building-a-chatbot-with-alter-nlu-f87007f27138?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/f87007f27138</guid>
            <category><![CDATA[nlu]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Wed, 17 Apr 2019 06:35:32 GMT</pubDate>
            <atom:updated>2019-04-17T06:45:34.639Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>Training a Chatbot on </strong>Alter NLU</h3><blockquote>The target audience for this post is developers who are looking for an open source platform meant for training or building AI-based conversational agents.</blockquote><p>In this post, I will be covering how you can build a simple dataset for your e-commerce chatbot and will point out those smartly handled pain points by our engineers which will make your dataset robust and complete.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FUidZjAAV_7gesMY0WbtpA.png" /></figure><h4><strong>Building no keyword Intents (Intents with no entities):</strong></h4><p>The intents like “<em>greet</em>” and “<em>exit</em>” are the generic intents every bot should handle. Besides these intents, build other intents you plan to include in your chatbot. Like, I have included “<strong><em>ongoing_offers</em></strong><em>”</em>.</p><p>This intent will hold all the user queries asking about the <em>current sales</em>, <em>vouchers</em> in our e-commerce chatbot.</p><p>Remember to train the dataset with expressions which contain words like — <em>sales</em>, <em>vouchers</em>, etc, as these words will keep the “<strong><em>ongoing_offers”</em></strong> intent unique from other non-keyword intents.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*0iL6yRsrx2rYXtTr" /><figcaption>Fig 1 : <strong><em>ongoing_offers </em></strong><em>intent example</em></figcaption></figure><h4><strong>Building Entities:</strong></h4><p>The e-commerce chatbot should be trained to handle queries like:</p><blockquote>User says : I want to buy <strong>apple</strong> <strong>mobile</strong> worth <strong>60K</strong></blockquote><p>So, I plan to create 3 entities that will extract the:</p><ul><li>brand</li><li>product-type and</li><li>price</li></ul><p>from the user query.</p><p>For each entity I add the ‘<em>Reference Value</em>’ and their synonyms. Like a user can write ‘<em>loui vuitton’</em> as ‘<em>LV’</em> or ‘<em>Louis</em> <em>Vuitton</em>’.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*admL8JMmoqqbyCpK" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/667/1*u9NKbaB-JVi8n74KuIZZnw.png" /><figcaption>Fig 2 : <em>Synonyms — Add here all the words you expect the user to say.</em></figcaption></figure><p>Similarly, create values and synonyms for the other fields.</p><h4><strong>Building Intents with Entities:</strong></h4><p>For queries as stated in the above section, dataset should have an intent that will store all the possible user queries from which the bot should be extracting the entities (like “<strong><em>search-product</em></strong><em>”</em> in this case). As stated in the image below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ce0b4dYFmGxKFMIv" /><figcaption>Fig 3 : <strong><em>search-product </em></strong><em>intent holding training phrases with entities.</em></figcaption></figure><p>For the developers ease we have built the console in such a manner that each ‘Selected Value’ can be linked to a ‘Reference Value’ of your choice.</p><p>Like in the images below, you can see in the intent section:</p><blockquote>Selected Value : 60k, 2k both have same ‘Reference Value’ i.e <strong>price-range</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/705/0*KeYY-VwRtKnEKpbU" /><figcaption>Fig 4 : The Intent Section</figcaption></figure><blockquote>In the entity section, every <strong>price-range </strong>example is defined in the same “Reference Value”</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Xo2jKPWpci5G61x-" /><figcaption>Fig 5 : The Entity Section</figcaption></figure><blockquote>If you change the ‘Reference Value’ in the entity section the same will be reflected dynamically in the intent section, and vice versa.</blockquote><h4><strong>Analysing the loopholes in the dataset: The Report Section</strong></h4><p>Once you are done building the dataset, move to the <strong>Report Section</strong> which will analyse your dataset for all intents and entities in real-time and notify the errors and warnings that need to be addressed for the accuracy of the chatbot’s response.</p><p>Once you have rectified all the errors, you will be able to download the dataset JSON in both — the Alter NLU or the RASA format.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1002/0*c4NShcfJf3tsuLTG" /><figcaption>Fig 6 : Alter NLU allows you to download JSON in 2 formats — the Alter NLU &amp; the RASA format</figcaption></figure><blockquote><strong><em>If you are using RASA NLU, you can quickly create the dataset using Alter NLU Console and Download it in RASA NLU format. We have updated our console for hassle free data creation which is less prone to mistakes.</em></strong></blockquote><h3><strong>Alter NLU Updates : v1.0.0-beta</strong></h3><h4><strong>Intent Model</strong></h4><p>We have used <strong><em>Convolutional Neural Networks (CNN) based model</em></strong><em> </em>to capture the intent. Further, the use of <strong><em>custom validation algorithm </em></strong>and <strong><em>matthews correlation coefficient </em></strong>as accuracy metric makes the intent model robust.</p><blockquote><em>The user has to train the sentence for one of the synonyms and the remaining are handled by our console code.</em></blockquote><h4><strong>Entity Model</strong></h4><p>In this version we have replaced the previous Flashtext and FuzzyWuzzy based entity extraction method with a <strong>CRF based Entity Recognition model.</strong></p><h3><strong>Build Your Bot:</strong></h3><p>Go to Git Repository from the link below:</p><blockquote>https://github.com/Kontikilabs/alter-nlu/tree/v1.0.0-beta</blockquote><p>Next, go through the README.MD file and start executing the steps as mentioned.</p><h4><strong>Below is an example along with a detailed explanation of the benefits of using this new pipeline.</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/457/0*avcGRVedmD6K0O4A" /><figcaption>Fig 7 : Output JSON</figcaption></figure><h4><strong>Elaboration of the above chatbot response:</strong></h4><ol><li>According to the context of the user query, the model successfully recognises the search product intent along with the confidence score.</li><li>This model handles <em>out-of-vocabulary </em>words to some extent. <br>The term ‘<em>out-of-vocabulary words</em>’ refers to those words which are not present in the training data of the chatbot.<br>For example, If you take a look at the example above, the <strong><em>parsed_value</em></strong> <em>“799k”</em> is not present in the training data used to train the “<em>ecomm-bot</em>” whose entity has been recognised accurately as “<em>price</em>”.</li><li>The CRF model was able to recognise the entity accurately, because it considers the sentence structure of the user query.</li><li>If you’re familiar with other bot frameworks, then you might not have come across a key like “<strong><em>parsed_value</em></strong>”. The main goal to add this key in the response is to assist developers to directly use the “<strong><em>parsed_value</em></strong>” if needed.<br>In the example above, the developer might need the exact value of entities such as “<em>price</em>” that is in the user query for further usage. In this case it’s “<em>799k</em>”.</li><li>Also, if you’re an existing user of Alter NLU it needs to be pointed out that the “<em>category</em>” key in the response has been renamed to “<em>name</em>”.</li></ol><h3>GIVE IT A SPIN!<br> <br>We’d love to hear all about your experience using Alter NLU. Critique is welcome too! We hope that using Alter NLU helps you create more effective and intelligent chatbots.</h3><blockquote><a href="https://console.kontikilabs.com/"><strong><em>Sign up on the console</em></strong></a><strong><em>.</em></strong></blockquote><blockquote><a href="https://kontikilabs.com/alter-nlu.html"><strong><em>Visit our website</em></strong></a><strong><em>.</em></strong></blockquote><blockquote><strong><em>For any further query, drop me an email on tanya[at]kontikilabs[dot]com.</em></strong></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f87007f27138" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/demo-on-building-a-chatbot-with-alter-nlu-f87007f27138">Demo on Building a Chatbot with Alter NLU</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How different and effective it is to train your chatbot using ALTER NLU]]></title>
            <link>https://medium.com/kontikilabs/how-different-and-effective-it-is-to-train-your-chatbot-using-alter-nlu-e35748d80700?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/e35748d80700</guid>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[chatbot-training]]></category>
            <category><![CDATA[naturallanguageprocessing]]></category>
            <category><![CDATA[nlu]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Fri, 22 Feb 2019 09:59:44 GMT</pubDate>
            <atom:updated>2019-02-22T13:19:37.832Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://console.kontikilabs.com/"><strong>Alter NLU</strong></a><strong> </strong>is designed to handle multiple chatbot datasets within a single user login. That means, one can conveniently add training data for multiple chatbots.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FNUCHYGZ2x9Mk9I0TqtAZg.png" /></figure><p>The main focus of Alter NLU is to get rid of the struggles involved in building a stable and good quality training dataset. To achieve this, we’ve segregated it into 3 parts:</p><ol><li><strong><em>Intents</em></strong></li><li><strong><em>Entities</em></strong></li><li><strong><em>Reports on training dataset quality</em></strong></li></ol><blockquote><strong>1. Intents</strong> help you recognise what your users want to say.</blockquote><p>There are 2 types of intents:</p><ul><li>Intents that do not map to any keyword, but are simple and direct phrases. A good example would be an intent named<strong><em> “greet”</em></strong> which will hold training phrases such as <em>“Hey there”</em> or <em>“Hi”</em>.</li><li>Intents that can have multiple or single keywords (entities) in the training phrase, and that we want to extract from the user query, as explained in <em>Fig. 1.1</em>.’s</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/443/1*XUuQlqO5gSNTMRfsFRzN8A.png" /><figcaption>Fig. 1.1</figcaption></figure><blockquote><strong>2. Entities</strong> describe the piece of information you would want to extract from the expressions/messages of the user. It identifies ‘things’ that your users have mentioned.</blockquote><p>For example in <em>Fig.</em> <em>1.1</em>, the <strong><em>brand_name</em></strong> and <strong><em>processor_gen</em></strong> are 2 different entities in our console that will be mapped to their respective <em>“Reference Value”</em> and <em>“Synonym”</em> values.</p><h3><strong>How Intents and Entities work to produce maximum output with relatively less training data?</strong></h3><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2Fr2BtghAUTmpP2%2Ftwitter%2Fiframe&amp;url=https%3A%2F%2Fgiphy.com%2Fgifs%2Fhigh-five-dancing-super-amazing-shit-r2BtghAUTmpP2&amp;image=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2Fr2BtghAUTmpP2%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="337" frameborder="0" scrolling="no"><a href="https://medium.com/media/7550da08b3b11d4a8a728d98570d4aea/href">https://medium.com/media/7550da08b3b11d4a8a728d98570d4aea/href</a></iframe><p><strong>The intent-sentence ratio &amp; managing an inventory of synonyms with entities<br></strong>The primary obstacle in building an intelligent chatbot is interpreting the varied patterns and expressions that a user may adopt when posing a query and training it for all possible approaches.</p><p>Unlike a few other NLP applications for training bots, we try to provide optimum results even when there are fewer training samples. <em>However, as a rule of thumb, the more you train your bot, the better the outcome.</em></p><blockquote>For good results, the console requires a minimum of 3 training samples per intent. Also, Alter NLU<strong> </strong>needs only a single synonym training sample for each entity reference value.</blockquote><p>To ensure that the training stats are appropriate, we show alerts and warnings in the reports section of the console.</p><p>For example, if a user wants to buy a laptop with certain specifications, he or she may pose a query like:</p><pre>           I want to buy a <strong>laptop</strong> with <strong>7th Gen</strong> processor</pre><p>The same query may be framed differently like:</p><pre>      Get me a <strong>PC</strong> embedded with a <strong>Seven Generation</strong> Processor.</pre><p>Our console allows you to maintain an inventory of related words such as “<em>laptop</em>” and “<em>PC</em>” in the form of entities, <strong><em>and train it only for a single synonym</em></strong>. This is illustrated in <em>Fig. 1.2</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wWGQ5rPG-DiA00rJ8JJAew.png" /><figcaption>Fig. 1.2</figcaption></figure><p><em>So, all you need to do is add one expression in the intent containing any one of the synonyms and your bot is good to go.</em></p><p><strong>Using Natural Language Processing for entity extraction<br></strong>Our engineers have incorporated <em>Natural Language Processing </em>(NLP) for entity extraction and this makes the bot intelligent enough to handle sentences that carry split entity synonyms.</p><p>For instance, you have added the entity synonym “<em>mobile</em> <em>cover</em>” and trained your bot accordingly, but the user frames the query by splitting the words apart as follows:</p><pre>                  need <strong>cover</strong> for my <strong>mobile</strong></pre><p>With <strong>Alter NLU</strong>, your bot will be able to extract the correct entity and accurately understand what information the user needs.</p><blockquote><strong>3. Reports: </strong>To analyse the loopholes in your dataset</blockquote><p>The Alter NLU console doesn’t stop at helping train data; it also provides AI-based real-time ‘<em>Reports</em>’ listing out issues and recommending improvements.</p><blockquote>This section provides in-depth analysis and alerts for intents and entities that need more training and identifies specific points that need to be addressed. This helps maximise the accuracy of the chatbot’s response.</blockquote><p>Chatbot training is an ongoing process that should get better at every successive stage. With each improvement, the trainer/developer should have a clearer understanding of the changes made.</p><p>Here are the list of items that get covered under Reports:</p><ul><li><strong>Intent Distribution -<br></strong>It represents that number of intents created as shown in <em>fig 1.3</em>. It also represents the numerical proportions for the number of sentences present in each of the intents(count) and their respective percentages(count percent).</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SM9LS6C_nol_DOav67hZBg.png" /><figcaption>Fig 1.3</figcaption></figure><ul><li><strong>Intents that require more training sentences -<br></strong>It reports the specific intents that have less number of training sentences than the <strong><em>threshold set i.e 3 per intent</em></strong>. Alerts the user with the name of the intents lacking enough training expressions as compared to the other intents in the bot.s</li><li><strong>Limitations in the entity section -<br></strong>Reports about the name of the entities which have been defined, but the user might not have formed any training sentence with it in the intent section. The other reason could be that the user might have mistakenly deleted the entity from the intent section but forgot to delete the same from the entity section.</li><li><strong>Examines the training dataset to extract the untagged entities</strong> - <br>Lists out keywords which have been tagged as an entity in intent but, the same keyword also occurs untagged in the training sentence of another intent.<br>Notifies the user that they might have skipped tagging the keyword as an entity in the other intent mentioned.</li><li><strong>Captures repetition of training sentence -<br></strong>Informs about the training sentence(s) which the user might have added in multiple intents by mistake. The console alerts this to the user with an error message at the top of the reports section.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vjcuhRpFL5RJs0PpsS80vA.png" /><figcaption>A screengrab of the console</figcaption></figure><h3><strong>How does it do? — The Engineering Involved</strong></h3><p>The background engineering involves identifying the intents and extracting entities to give an apt reply to the user query.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n_IJJNVDBNFXZfZnxgX-bg.png" /><figcaption>The background process</figcaption></figure><p><strong>The Intent Model<br></strong>Use of <em>cross-validation </em>and <em>matthews correlation coefficient </em>as accuracy metric makes the intent model robust. Further, we have used the <em>Convolutional Neural Networks (CNN) deep learning model </em>to capture the intent.</p><p>To attain better accuracy our engineers have implemented the preprocessing of training data in 2 steps:</p><p><em>1. Sentence Normalisation :</em><strong><br></strong>Sentence Normalisation helps in focusing on relevant parts of the text by discarding the irrelevant ones.</p><pre><em>         </em>For example, if the user says:<em> I </em><strong><em>need</em></strong><em> a </em><strong><em>laptop</em></strong></pre><p>Our code concentrates and extract the keywords “<em>need</em>” , “<em>laptop</em>” and discard the remaining unwanted text.<br>Similarly, we discard plurals from the text message.</p><p><em>2. Data Augmentation :</em><strong><br></strong>With the use of data augmentation, we make sure that the user does not have to explicitly add all the entity synonyms in the intent training section.</p><blockquote><em>The user has to train the sentence for one of the synonyms and the remaining are handled by our console code.</em></blockquote><p><strong>The Entity Model<br></strong>We have used both the <em>Flashtext</em>, a dictionary-based string match library which is a lot faster than the regex, and, <em>Fuzzywuzzy</em>, Levenshtein distance based string match library to extract all the entities, provided that the training data has a rich collection of the entities.</p><p>For the entity processing unit we extract and add all the root forms of the entity i.e for the plural keyword like “<em>laptops</em>” we form both “<em>laptop</em>” and “<em>laptops</em>”, but this is not the case with the singular keyword.</p><p>If you want to build your Chatbot and integrate it with Alter NLU, you can clone our code from the Git repository and follow the instruction mentioned in the ‘<a href="https://github.com/Kontikilabs/alter-nlu"><strong>Readme.md</strong></a>’ file.</p><h3>GIVE IT A SPIN!<br> <br> We’d love to hear all about your experience using Alter NLU. Critique is welcome too! We hope that using Alter NLU helps you create more effective and intelligent chatbots.</h3><blockquote><a href="https://console.kontikilabs.com/"><strong><em>Sign up on the console</em></strong></a><strong><em>.</em></strong></blockquote><blockquote><a href="https://kontikilabs.com/alter-nlu.html"><strong><em>Visit our website</em></strong></a><strong><em>.</em></strong></blockquote><blockquote><strong><em>For any further query, drop me an email on tanya[at]kontikilabs[dot]com.</em></strong></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e35748d80700" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/how-different-and-effective-it-is-to-train-your-chatbot-using-alter-nlu-e35748d80700">How different and effective it is to train your chatbot using ALTER NLU</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I, the destroyer]]></title>
            <link>https://medium.com/@tanyathakur6/i-the-destroyer-5756b850d7d4?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/5756b850d7d4</guid>
            <category><![CDATA[climate-change]]></category>
            <category><![CDATA[environmental-issues]]></category>
            <category><![CDATA[poetry]]></category>
            <category><![CDATA[global-warming]]></category>
            <category><![CDATA[self-destruction]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Sun, 19 Aug 2018 13:18:06 GMT</pubDate>
            <atom:updated>2018-08-19T13:18:06.670Z</atom:updated>
            <content:encoded><![CDATA[<h4>Modernising as No Blue, No Green</h4><p><em>A beautiful dwelling festooned by my creator…<br>To nurture whom I was born on earth as a janitor…</em></p><p><em>In the showers of ’94, I enjoyed bright sunlight and refreshing green…<br>The Fall brought golden sheen making the planet look serene…</em></p><p><em>Built memories with granny behind the dusty windowpane…<br>Humming, songs of the season until turning insane…</em></p><p><em>As I grew, worldly pleasures hauled me towards the unnatural machine…<br>The Green Panther was asleep till 2018, and the seasonal beauty now is unseen…</em></p><p><em>Was offered a majestic land full of colourful flowers and colourless water…<br>The former parched and the later turned red as I became a devil slaughter…</em></p><p><em>The barren land is now grinning and repudiating seeds…<br>And I turn helpless recalling my atrocious deeds…</em></p><p><em>Long ago I subjugated the forest and performed a sin…<br>Became an imaginary conqueror by overlaying the king’s skin…</em></p><p><em>High-handed I reached home to note victory hours…<br>Overlooking the To &amp; Fro pendulum, waving the future reversal of powers…</em></p><p><em>Greed was less satisfied with the furry coat…<br>To slay more, I landed on a lethal boat…</em></p><p><em>Exhausted I kept rowing the craft fronting the squall…<br>As from the shore, my mates had already eaten them all…</em></p><p><em>To master the ocean, I struggled to catch the orange roughy…<br>Ignoring my action may drastically hurt the geography…</em></p><p><em>Proudly I yelled VICTORY, clinching a suffocating life who could have survived for 150 years…<br>Tiny doppelgängers feared no new friend until 20 and swam away with invisible tears…</em></p><p><em>A major unknown damage was already done…<br>My triumph contributed one synthetic to existing millions of ton…</em></p><p><em>I devoured not only an orange angel but also an endangered blue whale…<br>Poor being swallowed the plastic and swam to the graveyard donning a white veil…</em></p><p><em>With time winters turned moody and summers became even hotter…<br>The spigot began to refuse my quench for water…</em></p><p><em>To butcher the Polar Bear I forced the sea ice to melt earlier…<br>My cruelty traveled miles making the situation even murkier…</em></p><p><em>Ere a minuscule and now a fervour destroyer of nature…<br>Slowly on earth, I was entitled as the only living creature…</em></p><p><em>Silence all over, I hear pendulum’s swaying sound, it is gods reversal of power recitation…<br>weak-kneed I swirl to see no woods but deforestation…</em></p><p><em>My doings transformed the only living planet into an abattoir...<br>Against nature, I have initiated an eternal war…<br>I have initiated an eternal war…</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5756b850d7d4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to build a clever automated travel companion ]]></title>
            <link>https://medium.com/kontikilabs/how-to-build-a-clever-automated-travel-companion-816e297e59af?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/816e297e59af</guid>
            <category><![CDATA[travel-bot]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[user-personas]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Wed, 20 Jun 2018 07:26:16 GMT</pubDate>
            <atom:updated>2018-06-20T07:26:16.231Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5AHSWYKd_tnC_W7vOmA5tA.jpeg" /></figure><h4>The inception of an AI-driven bot that goes way beyond just providing flight and hotel information</h4><p>Escapist, learner, planner, dreamer — no matter what you call yourself, if you are someone who travels extensively, it’s likely that you will not have the time to search the web and build an itinerary. Travellers rarely waste time in planning.</p><blockquote>“Over time, people will say, ‘I want to go to London,’ and the device will do everything, and you won’t have to do anything.” says — Priceline Executive Vice President, Glenn Fogel</blockquote><p>To make travel a smooth experience for such people, we, at <a href="https://kontikilabs.com/"><strong><em>KontikiLabs</em></strong></a>, promote the integration of Artificial Intelligence (AI) during the development of a travel bot so that travellers don’t have to go through some of the hassles of planning their trip. Instead, they can just pack their bags and fly, because:</p><blockquote>AI is not just about clicking buttons but, in a real sense, it is about inferring the social presence or the character of the person who is conversing with the bot. That is, AI should track the user persona.</blockquote><p>In the section below, are a few ideas, which if integrated intelligently, can help you construct a smart and an interactive user experience.</p><h3>How to develop a user persona? 🔎</h3><p>A high-level of <em>Machine Learning </em>(ML) and <em>Natural Language Processing </em>(NLP) can help in analysing and building the user profile based on social media likes.</p><blockquote><strong><em>The idea is:</em></strong><em> </em>If a user, via his Facebook profile likes an image <em>❤</em> of the Eiffel Tower, the bot should be able to discover that he may wish to travel to Paris. And, from the bot side, the user will be presented with ‘<strong>Travel to Paris’ </strong>option the next time he lands on the bot.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/865/1*18EiyFo1gzDGD1hxXchPWw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*RGWQ1vt3Ip6fx0Xz45k6Nw.png" /><figcaption><strong>Eg: With a Facebook like the bot will be able to intelligently learn and build user’s travel profile when he will click on — Fetch from FB</strong></figcaption></figure><h3>Contextual information — the intelligence level 🤓</h3><p>An intelligent and interactive travel bot should not only allow the user to enter different (travel-related) preferences, but should also <strong>remember the user profile, including his past travel preferences, last time travelled and other details.</strong></p><blockquote>So, if my previous stay was in a 3-star hotel, the bot should consider this and confirm with the user whether they would prefer a similar rating this time around too, before creating a new hotel preference profile in natural language.</blockquote><blockquote><a href="https://www.techradar.com/news/how-ai-is-changing-how-we-travel">Almost a third (29%) of global travellers say they are comfortable letting a computer plan an upcoming trip based on data from their previous travel history</a> and half (50%) don’t mind if they deal with a real person or computer, so long as their questions are answered.</blockquote><p>The bot should also be trained to fetch the last travel and stay details of the user when asked.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/456/1*wSHyGMXxZKFrNIvDm9WmIA.png" /><figcaption><strong>Fetching the last stay details and asking an intelligent question according the context.</strong></figcaption></figure><h3>How plug and play works — plan, book, and fly ✈️</h3><p>To build an effortless, yet swift user journey, it becomes necessary to <strong>complete the entire journey circuit on the same platform.</strong></p><p>Instead of going to a website and browsing options from different menus and filters, a bot can be powered by a collection of commands that allow the user to converse and plan his entire journey.</p><blockquote>The bot journey should be an engaging experience allowing the user to:<br> - Plan drafts for later use <br>- Fetch details about the last travel<br>- Build complex itineraries<br>- Personalise travel and hotel recommendations<br>- Make payments easily</blockquote><p>The moment a user shares their geo-location, the bot should become their companion. Considering the user persona-based <strong>personalised travel destination</strong> or any other location picked, a hotel stay option of the user’s choice should be presented as an option. <br>The user query may pertain to:<br>- price query, <br>- time query, <br>- preferences query</p><p>On opting the desired destination, the user <strong>should be able to complete the entire payment process</strong> either by scanning their passport or a saved passport, or even sharing details about a new credit card through the mobile app and make payment and identification responses. The bot’s job should not end here. When the bookings are done, a <strong>receipt of payment and travel details should be sent as an email to the user.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/714/1*0YFl3fbZkbeKpSVuZC3f9w.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/705/1*dPBM3Mj4VHIpdHQ-w2_WhQ.png" /><figcaption><strong>Fetch user persona and last user preferences to save time</strong></figcaption></figure><blockquote>“We see Artificial Intelligence as an evolutionary part of travel. We’ve always believed that people would go from click-type-tap style searching to a conversational format” : <a href="https://www.theguardian.com/sustainable-business/2017/feb/17/holidays-travel-automated-lastminute-expedia-skyscanner">says a Skyscanner spokeswoman</a></blockquote><h3>Planning a Group Trip — Because solo travel may not fascinate everyone👫👭</h3><p>Most people find it easier to travel with friends, who are likely to share in the responsibility of planning trips, than with older family members, who generally prefer to stay out of most of the responsibilities of trip planning. We understand that this can be a challenge.</p><p>Which is why, to make travelling with parents easy, we believe that the bot should allow the user to include others in the chat to form a group conversation.</p><blockquote>So, the user just adds members and the bot will automatically broadcast the plan for others to indicate acceptance or approval.</blockquote><h3>Attractive add-ons 🎯</h3><p>The focus should be on building an extensive warehouse of information that supports the user and makes his journey an amazing experience. To accomplish this, the bot should intelligently switch roles and guide the user through the following:</p><p><strong>Alerts! Now you can avail a </strong>🎫<strong> ticket just for $500 </strong>—The user should be able to set alerts and get a timely notification in natural language for airfare, hotel prices or any events specific to the trip.</p><blockquote>By adopting personal push notifications, airlines are seeing a 9% increase in conversion rates : <a href="http://www.webintravel.com/airlines-told-look-east-china-south-east-asia-drive-distribution-changes/">Travelport</a></blockquote><p>💵<strong> Dollar or </strong>💷<strong> Pound? Which currency to carry</strong> — The bot should be smart enough to infer that the travel is international, and provide information about the currency that the traveller should carry, based on the destination opted.</p><p>🗽 <strong>Places of interest </strong>— Filtering out popular tourist attractions may save a lot of time and manage the route to nearby places.</p><p>🌤 <strong>Weather Forecast — A few clouds</strong>:<strong> 4° C , Feels like 1° C — <br></strong>The bot should be able to forecast the weather for an area to help the user decide which place needs a jacket and which one will require a sunscreen. That is, it should offer pop-up suggestions that can help travellers decide what to carry on a trip.</p><h3>Availability — It travels with you 🛵</h3><p>The journey should not end post booking. A traveller should get insights into the places while travelling. The user should have the option to get <strong>real-time travel news feeds from various sources</strong>, filtered according to preference and rules, modifiable by the bot.</p><h3>Conclusion</h3><p>Chatbots are proving to create a big boom in the travel industry as they have the capability to provide interactive experience.</p><blockquote>Nearly 85 percent of travel and hospitality professionals are using AI within their businesses: <a href="https://www.iol.co.za/business-report/international/artificial-intelligence-to-make-travel-smarter-9935330">Tata Consultancy Services</a></blockquote><p>We, at <a href="https://kontikilabs.com/">KontikiLabs</a>, understand the real essence of AI and identify how to strategically leverage the technology along with NLP, at the right time in the customer’s journey. We promote and apply methods that leave our end user with a high-level of satisfaction, as it magnifies their sales and marketing objectives.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=816e297e59af" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/how-to-build-a-clever-automated-travel-companion-816e297e59af">How to build a clever automated travel companion 🚘</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine Learning 101 for Startups — Part 4]]></title>
            <link>https://medium.com/kontikilabs/machine-learning-101-for-startups-part-5-37d73173a58a?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/37d73173a58a</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[business-intelligence]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[business-strategy]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Fri, 09 Mar 2018 05:38:18 GMT</pubDate>
            <atom:updated>2018-05-23T11:29:59.523Z</atom:updated>
            <content:encoded><![CDATA[<h4>ML Visualisation and Data Analysis based on Amazon QnA Dataset</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jwWAAmcYvUjFMXv4oliVnw.jpeg" /></figure><blockquote>This is the fourth post in a series of posts on ML techniques for startups and business to help in decision making. The target audience for this post is business folks looking to understand how they can leverage Machine Learning in their business or product.</blockquote><p>Today, data is the game changer, and if used properly, data leads to better decision making. The quantity and quality of the data available are numerous, therefore, helping us in building new insights and making predictions, about what can and what will happen in the future.</p><p>We, at <a href="http://kontikilabs.com/">Kontiki Labs</a>, know the importance of data and what machine learning actions like entity extraction, clustering and more should be performed on it to help your business grow.</p><blockquote>The aim is to gain insights and extract relevant topics from data to detect what your customers are saying about your products.</blockquote><h4><strong><em>What is Machine Learning in the real world and, what can be done using it?</em></strong></h4><p>Machine Learning is the developed approach levied upon our computing methods to help our enterprise grow.</p><p>In this post, we will see how we can extract the maximum out of the <strong><em>Amazon QnA</em> </strong>dataset using visualisation tools powered by <strong>KNN</strong> and <strong>Clustering</strong> and benefit our business.</p><blockquote><strong>Note</strong>: We have tried our level best to dive in deeply into the answered questions for electronic items by the Amazon’s customers until the year of 2016. Our only limitation here is the permission to access the Amazon ASINs (ID of the product), but this does not effects the output retrieved and the positive impact our ML algorithm can pass onto your business.</blockquote><h4>The Amazon Data set</h4><p>This <a href="http://jmcauley.ucsd.edu/data/amazon/qa/"><strong><em>Amazon question/answer dataset</em></strong></a><strong><em> </em></strong>contains an enormous set of Question and Answer data from Amazon, figuring around 1.4 million answered questions. The set includes:</p><ol><li><strong>ASIN</strong> — ID of the product,</li><li><strong>questionType</strong> — type of questions, which can be ‘yes/no’ or ‘open-ended’,</li><li><strong>answerType</strong> — type of answer. Could be ‘Y’, ’N’, or ‘?’ (if the polarity of the answer could not be predicted). Only present for yes/no questions.</li><li><strong>question</strong> — question text.</li><li><strong>answer</strong> — answer text.</li></ol><p>In the next steps, you will see how our developers have used ML algorithm on the electronic product data and reconstructed it into various formats for different business use-cases.</p><h4><strong>The Data Cleanup</strong></h4><p>The data preparation process involves manipulating the ZIP file to JSON and then to CSV.</p><pre><strong><em>                         ZIP  → JSON → CSV                            </em></strong></pre><p>Once the ZIP file gets converted to a JSON format, we smartly append the data to a CSV file, which discarded storing it into the RAM. Therefore, saving our local system from massive dataset.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e5GL8kA8u6Wb56Pn9Qtwhw.png" /><figcaption><strong>Extracting text from ‘Read More’ till ‘Show Less’</strong></figcaption></figure><p>The data cleanup process also required cleaning special characters from customers queries and answers, like replacing semi-colons, smileys and more.</p><h3>Running the Code <strong>— </strong>Output Analyses</h3><h4>Analyses 1 — The question set comparison</h4><p><strong><em>Why Compare?</em><br></strong>Product reviews and the user’s comment not only helps in building trust between your brand and the customers but also provides a long-term exposure to your store. Your new customer base depends upon the satisfaction level of the existing customer base, because:</p><blockquote>70% of customers consult reviews or ratings before making a final purchase. (<a href="http://www.peopleclaim.com/#the-review-of-reviews">PeopleClaim</a>)</blockquote><p>To extract the maximum from both the reviews and comments the organisation should know what type of queries do customers have and how the other customers like to response to them.</p><p><strong><em>What we did?</em><br></strong>To help you in understanding the customer mindset, we did extensive machine learning and data analysis on the 2 type of questions in the dataset, the <em>yes/no</em> and <em>open-ended</em>.</p><p><strong><em>The Conclusion</em><br></strong>The conclusion is the pie chart below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-VutHRS93elOaJRd4OF71g.png" /><figcaption><strong>The blues sector represents — <em>open-ended, while the orange </em>represents — yes/no</strong></figcaption></figure><blockquote>The blue sector that covers more than 75% of the circle indicates that the majority were interested in answering open-ended questions in comparison to a simple yes or no.</blockquote><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/f4dec773b43660330c762af916526dd2/href">https://medium.com/media/f4dec773b43660330c762af916526dd2/href</a></iframe><h4>Analyses 2 — Understanding Customer’s Sentiments</h4><p><strong><em>Why Compare?</em></strong><em><br></em>Building a business is a journey that should never end, and while covering this journey, is a responsibility of meeting customer’s commitment to delivering high-quality work. <br>A shopkeeper in a physical store has a clear understanding of what his customers want, as they present verbal queries and suggestions, but what about an online store?<br>To bring the same experience, we implemented <strong>Sentiment Analyses</strong> on all the questions and answers asked by Amazon’s customers.</p><p><strong><em>What we did?</em><br></strong>We implemented sentiment analyses on all questions and answers posted by users for Amazon electronic products. With this, we inferred the opinion of the writer on the specific topic and his polarity towards the product. The output helped us in learning the response to the product which can be — <em>positive, negative or neutral</em>.</p><p><strong><em>The Conclusion:</em><br></strong>The outcome of the analysis on question and answer set is in the pie chart below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/763/1*PYJ4OP5XqhFHzrvBMoz1cA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/796/1*D0rzLa1Twx-bi79pUPAP4g.png" /><figcaption><strong>Question Sentiment vs Answer Sentiment</strong></figcaption></figure><p><strong><em>The motivation:</em> </strong>On comparing blue sectors from both the charts above, we can discover that the maximum of the neutral questions had a positive response from the customers.<br><strong><em>The bandwidth allocation: </em></strong>The <em>‘very negative’</em> and <em>‘negative’</em> sector in the pie chart helps us in learning where the problem is. With this, a leader can easily make out and allot man-hours to the desired area and get better results in the future.<br><strong><em>The revenue generator:</em></strong> The <em>‘very positive’</em> and <em>‘positive’</em> sector can help the organisation in knowing what products are creating the real buzz in the market and how can they start dealing and investing more on them to increase the company sales.</p><h4>Analyses 3 — The Rating Predictions</h4><p><strong><em>Why Compare?</em><br></strong>Ratings not only help customers with their purchase decisions but, also drive more qualified buyers towards your product. Hence, boosting your sales by a significant amount.<br>The ads which appear online also play a significant role because they represent aggregated rating data for the product gathered from varied sources including third-party reviews aggregators, merchants, editorial sites, and consumers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/851/1*8u9-fcPeM-Jmi4YAINu8Nw.png" /></figure><p><strong><em>What we did?</em><br></strong>We ran our algorithm to see how many answers for electronic products fall in what rating bracket.</p><p><strong><em>The Conclusion</em><br></strong>The conclusion is the bar chart below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E1_-6RktRNoxUPyVhRx2GA.png" /><figcaption><strong>The product rating bar graph</strong></figcaption></figure><p>The Y-axis represents the total number of answers, while the X-axis represents product rating. So, the blue bar indicates that more than 50 thousand answers have product rating of [0.9,1.0].</p><blockquote>With the results, it becomes quite evident that which products are mass stealers. This result can help the <strong>content management team</strong>, as they will know in what area they can build their content and can target the right audience for company profits.</blockquote><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/f4dec773b43660330c762af916526dd2/href">https://medium.com/media/f4dec773b43660330c762af916526dd2/href</a></iframe><h4>Analyses 4 — Get the product report</h4><p><strong><em>Why Compare?</em></strong><em><br></em>In a highly ambitious global marketplace, the pressure on organisations to find new ways to deliver the best to their customers grows ever stronger. The increasing need for the industry to compete with its products in a global market across quality and service dimensions has given rise to the need to develop more dynamic inventory strategies. <br>Therefore, it becomes important for the organisation to analyse timely detection of shifts in demand, and know in which month or time slot customers show maximum interest in their product.</p><p><strong><em>What we did?</em><br></strong>Our ML study identified each month and then each day in a month to figure out the period involving maximum interest rate of the customers towards the products.</p><p><strong><em>The Conclusion</em><br></strong>The conclusion is the bar chart below, which shows the peak time when the customers were alert.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dsJqfyYMuqAykr5vyZp1vg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R53PfXRzUH08aN2sG0Q4yA.png" /><figcaption><strong>Monthly Comparison &amp; Comparing on annual basis</strong></figcaption></figure><blockquote>The result can prove to be a big boon for businesses as they know what time of the year they need maximum employee strength.<strong><br>From the managerial fronts</strong>, an HR or a support management team can judge in prior, the time in the year when the company is in need of mass hiring.</blockquote><p>As the graph points out January and December, which are progressing towards crossing the 10 thousand mark. This is a signal to the sales department, to react quickly to changes in sales demand or inventory stocks, and respond faster to meet customer requirements.</p><h4>The code gist</h4><p>Below is an example prepared by our developers. The code will help you in collecting all the results explained above.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/59d1581b6ff6f8ad221012e1defe5cd6/href">https://medium.com/media/59d1581b6ff6f8ad221012e1defe5cd6/href</a></iframe><h3>Conclusion</h3><p>The study identified some important aspects that machine learning and data analysis can bring to build an efficient business. It suggests how we can loop into our existing dataset and derive ways that can be progressive and customer-centric. And to integrate the same experience in your business, we, at <a href="http://kontikilabs.com/">Kontiki Labs</a>, have provided the code above for ML Visualisation based on Amazon QA Dataset that you can leverage to monitor your business stats.</p><blockquote><em>At </em><a href="http://kontikilabs.com/"><em>Kontikilabs</em></a><em>, we help our customers build such custom Machine Learning and deep learning tools, with a focus on the data available, the problems to be solved and the end user objectives. You can connect with us on </em><a href="https://twitter.com/kontikilabs"><em>twitter</em></a><em> or on email: hi[at]kontikilabs[dot]com</em></blockquote><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/f4dec773b43660330c762af916526dd2/href">https://medium.com/media/f4dec773b43660330c762af916526dd2/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=37d73173a58a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/machine-learning-101-for-startups-part-5-37d73173a58a">Machine Learning 101 for Startups — Part 4</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Comparing Machine Learning (ML) Services from Various Cloud ML Service Providers]]></title>
            <link>https://medium.com/kontikilabs/comparing-machine-learning-ml-services-from-various-cloud-ml-service-providers-63c8a2626cb6?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/63c8a2626cb6</guid>
            <category><![CDATA[amazon]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[ibm-watson]]></category>
            <category><![CDATA[microsoft-azure]]></category>
            <category><![CDATA[google]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Tue, 23 Jan 2018 06:32:19 GMT</pubDate>
            <atom:updated>2018-05-23T11:39:47.822Z</atom:updated>
            <content:encoded><![CDATA[<p>The ongoing hustle in the AI world is demonstrating that integrating ML/Natural Language Processing (NLP) in nearly any area, even making a machine and human converse via voice, is no rocket science. In fact, the technology/AI application enabling this — chat bots/voice bots — is fast evolving.</p><p>The launch of Amazon Echo and Google Home has shown the promise of voice, especially to businesses. This is giving us the impetus to start planning for and building voice bots. Interestingly, the arrival of Deep Learning technologies like Automatic Speech Recognition (ASR) and NLP from tech giants like Amazon and Google, has made the technology accessible.</p><p><strong>These companies provide ML services, which help in:<br>●</strong> Extracting keywords from words to understand what you are writing.<br>● Scanning and displaying the elements in an uploaded image or video.<br>● Converting text to speech and vice-versa.</p><p>We, at <a href="http://kontikilabs.com/">Kontiki Labs</a>, understand that there are several service providers and services available for NLP, vision APIs and text &amp; speech, making it difficult to decide which will be right for your business. So, we have looked through them and created a ready-reckoner that will help you make the right choice based on your business need. We have considered many crucial factors to create this quick guide which you can leverage to make a better product generation:</p><p>These are the Machine Learning services that we are going to cover:<br>1. Natural Language Processing services<br>2. The Vision APIs<br>3. Text-to-Speech &amp; Speech-to-Text services</p><h3>1. Natural Language Processing Services</h3><h4>Amazon Comprehend NLP vs. Google Cloud Natural Language vs. IBM Watson NLU vs. Microsoft Azure Text Analytics</h4><p>When building a chatbot, there are numerous factors to consider. The first among them is: How will the machine know what the user is writing?</p><p><strong>The Solution: <br></strong>Leverage Amazon Comprehend, Google Cloud Natural Language and services from different providers that provide easy-to-use APIs. These NLP services use ML to extract insights from the text entered by the user.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/491/1*Z2wbQUsGw_aLUYVnDudEog.png" /><figcaption><strong>How important keywords are extracted from the text.</strong></figcaption></figure><p>Below are the 3 key aspects that we have considered to compare the NLP services — features, code execution and output, and price:</p><h4>i. Feature Comparison</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*_OflhJ50fbNf2KE1xjddag.png" /><figcaption><strong>Feature Comparison of NLP services</strong></figcaption></figure><h4>ii. Code Execution &amp; Output Comparison</h4><p>The idea is to pass same text to all the NLP services and see which one provides us with the most relevant keywords, as output.</p><p><strong><em>Amazon Comprehend: <br></em></strong>Below is a very simple example prepared by our developers. The code will help you in executing the Amazon comprehend ML service:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8e4e0e4e96fecff596aab91bfd037bbc/href">https://medium.com/media/8e4e0e4e96fecff596aab91bfd037bbc/href</a></iframe><p><strong><em>Google Cloud Natural Language: <br></em></strong>Execute the code below to know how Google NLP service responds to the same text you passed for Amazon above.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/9b4657228f5d289da550bfd520371c55/href">https://medium.com/media/9b4657228f5d289da550bfd520371c55/href</a></iframe><p><strong><em>IBM Watson NLU:</em></strong><em><br></em>The code below will reveal the features provided by the IBM Watson NLP service</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/ef9e2fcf4c69b778f56c45f7605c753b/href">https://medium.com/media/ef9e2fcf4c69b778f56c45f7605c753b/href</a></iframe><p><strong><em>Microsoft Azure Text Analytics:</em></strong><em><br></em>Below is the code for executing the Microsoft Text Analytic service.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/34a14e2f8c515dec1250d427eaeb4bf3/href">https://medium.com/media/34a14e2f8c515dec1250d427eaeb4bf3/href</a></iframe><p>The output comparison for all the NLP service providers :</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/547/1*r15pJYSbGLjLKPF4vA6WXw.png" /><figcaption><strong>Time taken by the NLP services to execute the features provided.</strong></figcaption></figure><h4>iii. Price Comparison of NLP Services</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*bYjHJqMKgaTY_kmdKZkC8w.png" /><figcaption><strong>Comparing price plans of NLP services</strong></figcaption></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/f4dec773b43660330c762af916526dd2/href">https://medium.com/media/f4dec773b43660330c762af916526dd2/href</a></iframe><h3>2. Vision APIs : <strong>Moving beyond text</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/390/1*GEmU7gRa5D2y_1AwNRwo0w.png" /><figcaption><strong>A bot that extracts text and tags from an image.</strong></figcaption></figure><h4>Amazon Rekognition vs. Google Vision vs. IBM Watson Visual Recognition vs. Azure Computer Vision API</h4><p>The services allow you to add image and video analysis to your bot/application. The easy-to-use APIs return the content in the images or videos uploaded.</p><p>Below are the 3 key aspects that we have considered to compare the vision API services — features, code execution and output, and price:</p><h4>i. Feature Comparison</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*QR8zjN0DloS-YxCmrz__qQ.png" /><figcaption><strong>Feature Comparison of Vision services</strong></figcaption></figure><h4>ii. Code Execution &amp; Output Comparison</h4><p>We shared a single image with all the service providers API and compared the image recognition results.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/715/1*7DVXvGkE7u8iGHBmG0YqWw.jpeg" /><figcaption><strong>We opted this image as this can extract maximum features : Facial Recognition, landmark and text extraction.</strong></figcaption></figure><p><strong><em>Amazon Rekognition:</em></strong><em><br></em>Execute the code below to see how Amazon Rekognition service scans the image above.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/7661eebcbef0f8e67fe912105702be9a/href">https://medium.com/media/7661eebcbef0f8e67fe912105702be9a/href</a></iframe><p><strong><em>Google Vision:</em></strong><em><br></em>Below code is prepared by our team to see how Google Vision reads the image.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/dd0edd521e00f1dac2b414d804b66a9f/href">https://medium.com/media/dd0edd521e00f1dac2b414d804b66a9f/href</a></iframe><p><strong><em>Microsoft Computer Vision API :</em></strong><em><br></em>To analyse your image with Microsoft Azure use the sample code provided in the section below.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/cc59f7835d2dc6cd41dd2b3521d89a68/href">https://medium.com/media/cc59f7835d2dc6cd41dd2b3521d89a68/href</a></iframe><p>The output comparison for the Vision API :</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/547/1*VXujhExzrqdJlToM8nnf7A.png" /><figcaption><strong>Time taken by the vision services to execute the features provided.</strong></figcaption></figure><h4>iii. Price Comparison of Image Services</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*2sF6984kBuF5R6PWI5bukQ.png" /><figcaption><strong>Comparing price plan of image services</strong></figcaption></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/f4dec773b43660330c762af916526dd2/href">https://medium.com/media/f4dec773b43660330c762af916526dd2/href</a></iframe><h3><strong>3. Processing Text and Speech</strong></h3><p>Thinking of building a voice bot? Well, the API driven services from the below providers can become the backbone of your bot.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/576/1*ZP-O-VSzdV_28iWpB5ROqA.png" /><figcaption>A Shop bot example : perceives human voice and transforms it into text.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*8xCT163YV2zk5OwbmseHVA.png" /><figcaption><strong>Different provides and their services to process text</strong></figcaption></figure><h4>a. Text-to-Speech (TTS)</h4><p>Below are the 3 key aspects that we have considered to compare text-to-speech services — TTS feature analysis, code execution and output, and price:</p><h4>i. TTS Analysis</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*41R1LJYFLwesh-QSSH-TxQ.png" /><figcaption><strong>Comparing Text-to-Speech</strong></figcaption></figure><h4>ii. Code Execution &amp; Output Comparison</h4><p>We passed same text to all the speech API so that it becomes easy for you to figure out that which one provides clear output.</p><p><strong><em>Amazon Polly:<br></em></strong>The section below provides the code for executing text-to-speech via Amazon.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/eda8adbccd1344784a64e83d87e53ced/href">https://medium.com/media/eda8adbccd1344784a64e83d87e53ced/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fw.soundcloud.com%2Fplayer%2F%3Furl%3Dhttp%253A%252F%252Fapi.soundcloud.com%252Ftracks%252F384909869%26show_artwork%3Dtrue&amp;url=https%3A%2F%2Fsoundcloud.com%2Fuser-405181736%2Fpolly-stream-testtext&amp;image=http%3A%2F%2Fa1.sndcdn.com%2Fimages%2Ffb_placeholder.png%3F1515765262&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=soundcloud" width="800" height="166" frameborder="0" scrolling="no"><a href="https://medium.com/media/541cdbb779c73e718eaffe30017208e1/href">https://medium.com/media/541cdbb779c73e718eaffe30017208e1/href</a></iframe><p><strong><em>Microsoft Bing Speech API — Text-to-Speech:<br></em></strong>Convert your audio to text by using the below code<strong><em> :</em></strong></p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/a1ec93a376786c48abe6c4d68556cbb8/href">https://medium.com/media/a1ec93a376786c48abe6c4d68556cbb8/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fw.soundcloud.com%2Fplayer%2F%3Furl%3Dhttp%253A%252F%252Fapi.soundcloud.com%252Ftracks%252F384909941%26show_artwork%3Dtrue&amp;url=https%3A%2F%2Fsoundcloud.com%2Fuser-405181736%2Fbing-testtext-tts&amp;image=http%3A%2F%2Fa1.sndcdn.com%2Fimages%2Ffb_placeholder.png%3F1515765262&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=soundcloud" width="800" height="166" frameborder="0" scrolling="no"><a href="https://medium.com/media/f858b2061e814da50dde8bb2b7be0922/href">https://medium.com/media/f858b2061e814da50dde8bb2b7be0922/href</a></iframe><p><strong><em>IBM Watson Text-to-Speech:<br></em></strong>Convert your written text to voice by using the below code</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/396b4f5218a4b35cd2c234a63cb77ee0/href">https://medium.com/media/396b4f5218a4b35cd2c234a63cb77ee0/href</a></iframe><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fw.soundcloud.com%2Fplayer%2F%3Furl%3Dhttp%253A%252F%252Fapi.soundcloud.com%252Ftracks%252F384909983%26show_artwork%3Dtrue&amp;url=https%3A%2F%2Fsoundcloud.com%2Fuser-405181736%2Fwatson-testtext-tts&amp;image=http%3A%2F%2Fa1.sndcdn.com%2Fimages%2Ffb_placeholder.png%3F1515765262&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=soundcloud" width="800" height="166" frameborder="0" scrolling="no"><a href="https://medium.com/media/65af5f3adc20756f77a99b5fd0ec6f5f/href">https://medium.com/media/65af5f3adc20756f77a99b5fd0ec6f5f/href</a></iframe><h4>iii. Price Comparison for Text-to-Speech</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*C1x6SOsDJv1W5XCa4g9wuQ.png" /><figcaption><strong>Price comparison for text-to-speech</strong></figcaption></figure><h4>b. Speech-to-Text (STT)</h4><h4>i. STT Analysis</h4><p>Below are the 3 key aspects that we have considered to compare the speech-to-text services — STT feature analysis, code execution and output, and price:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*LFVFC3bbvljamKRQss7otQ.png" /><figcaption><strong>Comparing Speech-to-Text</strong></figcaption></figure><h4>ii. Code Execution &amp; Output Comparison</h4><p><strong><em>Google Cloud Speech API:</em></strong><em><br></em>Below is the code for executing the Google STT API</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/6981e97ba07585dcb143afdbbd15dcae/href">https://medium.com/media/6981e97ba07585dcb143afdbbd15dcae/href</a></iframe><p><strong><em>Microsoft Bing Speech API — Speech-to-Text:</em></strong><em><br></em>The section below will help you in converting your speech to text.</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/8899549eac72ac03c3fc18109366716e/href">https://medium.com/media/8899549eac72ac03c3fc18109366716e/href</a></iframe><p><strong><em>IBM Watson Speech-to-Text :</em></strong><em><br></em>Execute the code below to convert your speech to text using IBM Watson</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/e4a51bc1f01afcdc43e2d1c65ffbf807/href">https://medium.com/media/e4a51bc1f01afcdc43e2d1c65ffbf807/href</a></iframe><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/1*fQF-OPO9Dl2dFu6ZRFzv1g.png" /><figcaption><strong>Time taken by Speech-to-Text services to execute the features provided.</strong></figcaption></figure><h4>iii. The Price Comparison for Speech-to-Text</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/560/1*hNF5ah6ecGvA0e1H-fJzXw.png" /><figcaption><strong>Price comparison for speech-to-text</strong></figcaption></figure><h3>4. Conclusion</h3><p>While most ML service products have common features, there are plenty that make them unique. Prior to selecting one, you should consider the type of product that you want to build. And to facilitate this decision, we, at <br><a href="http://kontikilabs.com/">Kontiki Labs</a>, have provided the code above for ML services that you can leverage to monitor the available services and opt for one that fulfils all your needs.</p><blockquote>At <a href="http://kontikilabs.com/">Kontiki Labs</a>, we are enabling enterprises and businesses of all sizes to use AI powered technologies such as Machine Learning and Deep Learning by building affordable, people focused, design-first AI solutions. You can connect with us on <a href="https://twitter.com/kontikilabs">twitter</a> or on email: hi[at]kontikilabs[dot]com</blockquote><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/4c93781c4449ad62164da06fdbe9aeea/href">https://medium.com/media/4c93781c4449ad62164da06fdbe9aeea/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=63c8a2626cb6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/comparing-machine-learning-ml-services-from-various-cloud-ml-service-providers-63c8a2626cb6">Comparing Machine Learning (ML) Services from Various Cloud ML Service Providers</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Machine Learning 101 for Startups — Part 1]]></title>
            <link>https://medium.com/kontikilabs/machine-learning-101-for-startups-part-2-6084c80587d9?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/6084c80587d9</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[business-intelligence]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[logistic-regression]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Wed, 17 Jan 2018 08:34:20 GMT</pubDate>
            <atom:updated>2018-05-23T11:40:08.514Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JN_iq1m57Ai561qX-k7VbA.png" /></figure><blockquote>This is the first post in a series of posts on ML techniques for startups and business to help in decision making. The target audience for this post is business folks looking to understand how they can leverage Machine Learning in their business or product.</blockquote><p>How confusing it becomes when the output is even more complex than the problem statement? To take you out of this loop and help in your decisions we introduce you to the concept of Logistic regression. For instance:<br><strong><em>Question</em></strong><em>:</em> Do body weight, fat consumption, and age have an impact on the probability of having a heart attack?<br><em>The answer predicted by the algorithm is a simple yes or no, thus resulting in quick conclusions.</em></p><h3>The problem statement — How to classify new data based on existing data.</h3><p><strong>Logistic regression </strong>is generally used when the dependent variable is binary or Dichotomous.</p><p>That is the ‘<em>dependent variable’</em> can take only 2 possible values such as:<br><em>“Yes/No”</em>, <em>“Default/No Default”</em>, <em>“Living/Dead”</em>, <em>“Responder/Non Responder”, true/false </em>etc<em>, </em>while the ‘<em>independent factors’</em> or variables can be both categorical or numerical.</p><p>Logistic regression is a combination of regression and logistic function which predicts the probability of occurrence of an event. <strong>So, <em>if the probability is less than 0.5, then the predicted variable is 0, and if the probability is more than 0.5, then the predicted variable is 1.</em></strong></p><p>Even though logistic regression is frequently used for binary variables (2 classes), it can also be used to handle categorical dependent variables having more than two classes, which is termed as <strong>Multinomial Logistic Regression</strong>.</p><p><strong>A simple example of Multinomial Logistic Regression.<br></strong>Here we are considering the famous <strong>Iris flower data set.</strong> The dataset, introduced by the British statistician and biologist Ronald Fisher, consist of various flowers and their properties like the length and width of the sepals and petals, in centimetres. The data set comes preloaded in the <a href="http://scikit-learn.org/stable/"><strong>sklearn library.</strong></a></p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/50383b11b02146867609f2058722182f/href">https://medium.com/media/50383b11b02146867609f2058722182f/href</a></iframe><p>In the above code, we see that the logistic classifier accuracy is 86%, which proves to be pretty good for such a minute dataset.</p><p>The algorithm is advised when you want to follow a probabilistic approach in your product model. This method proves to be useful when in future you want to append and immediately incorporate more training data into your model. The algorithm is quick to train and classify unknown records.</p><p>It is important to note that in a situation where the training data is sparse and high dimensional, the logistic model may overfit the training data. <br>Therefore, for stable and meaningful outcomes, you need to provide Logistic regression algorithms with a huge dataset.</p><p>Like the algorithm above, we will come up with more Machine Learning tools, that will help you in understanding different scenarios and producing robust outputs.</p><p>At <a href="http://kontikilabs.com/">Kontikilabs</a>, we help our customers build such custom Machine Learning and deep learning tools, with a focus on the data available, the problems to be solved and the end user objectives. You can connect with us on <a href="https://twitter.com/kontikilabs">twitter</a> or on email: hi[at]kontikilabs[dot]com</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F66c993%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F66c993%2F&amp;image=https%3A%2F%2Fe.enpose.co%2F%3Fkey%3DdRXnS9Gplk%26w%3D700%26h%3D425%26url%3Dhttps%253A%252F%252Fupscri.be%252F66c993%252F%253Fenpose&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/f4dec773b43660330c762af916526dd2/href">https://medium.com/media/f4dec773b43660330c762af916526dd2/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6084c80587d9" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/machine-learning-101-for-startups-part-2-6084c80587d9">Machine Learning 101 for Startups — Part 1</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Those 22:03 mins. Bots, where are you?]]></title>
            <link>https://medium.com/kontikilabs/those-22-03-mins-bots-where-are-you-bc46ad4acaf3?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/bc46ad4acaf3</guid>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[kontiki-design]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[chatbots]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Wed, 10 Jan 2018 17:48:09 GMT</pubDate>
            <atom:updated>2018-01-12T23:08:31.106Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ePrXxOP9C7_tH8brGAabuw.jpeg" /></figure><h4><em>How bots in customer care service can end customer frustration.</em></h4><p><em>“Nothing can replace humans!” </em>🤓</p><p>I had always held onto this belief; that is, until the day I had to contact the customer care service of one of the most reputed banks to know the complaint status of an unauthorised transaction.</p><p>The call lasted for around 22:03 minutes, out of which, for approximately 17 minutes, I was forced to listen to one of the most annoying hold tones in the world. For 3 minutes, my call was tossed between multiple executives. Apparently, they have different departments for different queries.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LZ_OgJKVTzY_p6M_7j3vDA.png" /><figcaption>My Customer Care Call Journey 😔</figcaption></figure><p>What really put me off after all this, was the answer I received to the query I finally managed to convey after the long wait. As the reply to my questions about the transaction and complain status, came the million-dollar reply:</p><blockquote><em>‘Ma’am your request is under process…’. <br></em><strong>The result: My crucial </strong><em>⏰</em><strong> 22:03 minutes and my phone balance </strong><em>💰</em><strong> were spent in vain.</strong></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*9gceyUG4QlZMU1DXZDBGRA.gif" /></figure><p>It is not that this was my first time facing such a situation. But, considering that we are in 2018, and we have technologies which make things more approachable, interactive and efficient, why are we not using them? 🤷🏼‍</p><p>It is high time that we let our human minds be occupied with important tasks and replace the <strong><em>basic level customer service with the AI-driven bots </em></strong>🤖 .</p><p>My suggestion probably led you to come up with some of these responses:<br><em> 1. What are Bots </em>🤔<em> ?<br> 2. Are they powerful enough to replace humans?<br> 3. If yes, then this will cause unemployment.<br> 4. Oh, cm’on!</em>🙄<em> Nothing can replace humans.</em></p><p>To sum things up:<br> <strong><em>My definition of bots is:</em></strong><br> Bots enable humans to converse with a device 🤳 (computer/mobiles) and get instant replies.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/742/1*7LIkWanqwi6Jonto0lKV0Q.png" /></figure><p>If you want to dive in and explore more on this, do check out <a href="https://whatarechatbots.com/"><strong><em>What Are Chatbots</em></strong></a><strong><em>.</em></strong></p><p>To allay the remaining confusion, let us talk facts. The facts say: <br> <strong><em>Yes, bots are </em></strong>💪🏽<strong><em> powerful:</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ve2obZzaOvCIR86AecEVwQ.png" /><figcaption>The Bot Journey 😁</figcaption></figure><blockquote><a href="https://www.gartner.com/imagesrv/summits/docs/na/customer-360/C360_2011_brochure_FINAL.pdf"><em>Gartner projects </em></a><em>that more than 85 percent of customer interactions will be managed without a human by 2020.</em></blockquote><p><strong><em>Not unemployment; but enabling humans to apply their brains towards better solutions for the remaining 2/3 crucial tasks:</em></strong></p><blockquote><a href="https://www.gartner.com/newsroom/id/2956618"><em>Gartner believes </em></a><em>that by 2017, only one-third of customer service interactions will require human support.</em></blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/653/1*gFEgyAzB1W3pPjAXFKWAQw.png" /></figure><p>The last fact pertains to the <strong><em>NOTHING</em></strong><em> </em>in<em> ‘Nothing can replace humans’</em>:</p><blockquote><em>Twenty-two percent of consumers claim to have used chat apps to interact with banks for services such as account activation and balance check.</em></blockquote><h4><strong>‘Answers to customer’s trivial queries should be managed by bots’</strong></h4><p>Let us see how a bot solves a user query in seconds:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/393/1*OB_qDhkwsCO176tc1Vo1Tg.png" /><figcaption>A Support Bank Bot</figcaption></figure><p>A bot like a <strong>BankBot</strong> can easily reside on Facebook Messenger, your website, slack or any other medium to solve your customer’s query.<br> The user is simply required to send the complaint number.</p><p>With a high level of M<strong>achine Learning</strong> and <strong>Natural Language Processing </strong>(NLP), the bot studies the user’s text and sends the query to hit an API request whose response can be fetched within seconds, thus saving the user’s time.</p><p>Still wondering what your employees will do?</p><p>Well, they will focus their efforts and knowledge towards solving the customer’s problem, rather than just answering 😎</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bc46ad4acaf3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/kontikilabs/those-22-03-mins-bots-where-are-you-bc46ad4acaf3">Those 22:03 mins. Bots, where are you?</a> was originally published in <a href="https://medium.com/kontikilabs">Kontiki AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beasts of Burden, who?]]></title>
            <link>https://medium.com/@tanyathakur6/beasts-of-burden-who-694027dde11?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/694027dde11</guid>
            <category><![CDATA[dominance]]></category>
            <category><![CDATA[burden]]></category>
            <category><![CDATA[farmers]]></category>
            <category><![CDATA[animals]]></category>
            <category><![CDATA[suicide]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Fri, 22 Sep 2017 08:22:15 GMT</pubDate>
            <atom:updated>2017-10-16T11:04:11.668Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HCRE-uDX3Ed7Dxh0RnOyBQ.png" /></figure><p><em>Sweat all over, he is on a working spree</em><strong><em>…<br></em></strong><em>Tired eyes clashes with the sun to assume, the clock displaying three</em><strong><em>…</em></strong></p><p><em>Time for the food provider to eat</em><strong><em>…<br></em></strong><em>Sand on the hot ground is his infinite seat…</em></p><p><em>Worried about his wages, he gazes at his OX</em><strong><em>…<br></em></strong><em>Torn apart, the scene doesn’t let him unbox…</em></p><p><em>A bird performing vocalisation while sitting on the OX hump</em><strong><em>…<br></em></strong><em>Jaded OX turnaround, and the bird jumps…</em></p><p><em>The hungry black Drongo devoured, and flew away…</em><strong><em><br></em></strong><em>Unveiling to the farmer, his betray…</em></p><p><em>His muzzled OX could only smell..</em><strong><em><br></em></strong><em>A life, no less than hell…</em></p><p><em>He recalls the bible which says, ‘</em><strong><em>shall not muzzle an ox when it is treading out the grain</em></strong><em>’..</em><strong><em><br></em></strong><em>But, what about the invisible muzzle of the farmer, isn’t that inhumane?…</em></p><p><em>Ignoring the pittance that he earns…</em><strong><em><br></em></strong><em>To hug and unmuzzle his OX, he runs…</em></p><p><em>Both cry loud, it’s a cathartic release..</em><strong><em><br></em></strong><em>Not money, but their bond give them a tender breeze…</em></p><p><em>Loud the lord ask the world, for the good of this huddle...</em><strong><em><br></em></strong><em>Who is the </em><strong><em>beast </em></strong><em>behind</em><strong><em> </em></strong><em>their endless</em><strong><em> burden?</em></strong></p><p><em>Somewhere else a face lacking oomph, inside a broken shack, makes hour-long rain vigil…</em><strong><em><br></em></strong><em>Cursing himself for this day, because towards his camel the behaviour was evil…</em></p><p><em>The watery orbs continuously stares at his livestock…</em><strong><em><br></em></strong><em>Until, while grazing he looses his balance because of a rock…</em></p><p><em>Wailing sound from both ends as the camel falls off the mountain…</em><strong><em><br></em></strong><em>Not the clouds, but the farmer’s eyes have now plenty of rain…</em></p><p><em>Hearing the scream, his wife runs toward the hut leaving lugs of trees…</em><strong><em><br></em></strong><em>Seeing the visuals, tight-lipped she sits on her knees…</em></p><p><em>The two worried human eyes meet in silence, and talk about their abject poverty…</em><strong><em><br></em></strong><em>Alluding each other, loss of the only left wealth and security…</em></p><p><em>Crawling around and knocking doors to find a saviour…</em><strong><em><br></em></strong><em>Their helpless comrade says, the dominance gave him trifling 17 rupee as waiver…</em></p><p><em>High debt and the monsoon failure, forces them to end their existence…</em><strong><em><br></em></strong><em>And the dominance ploy the homicide, as suicide of innocence…</em></p><p><em>Loud the lord ask the world, for the good of this huddle…</em><strong><em><br></em></strong><em>Who is the </em><strong><em>beast </em></strong><em>behind</em><strong><em> </em></strong><em>their endless</em><strong><em> burden?</em></strong></p><p><em>Beasts are the elite, with mala fide intentions…</em><strong><em><br></em></strong><em>The 20% use the nations money to build their mansions…</em></p><p><em>You have all forced them to be docile…</em><strong><em><br></em></strong><em>You have all made a fecund land, infertile…</em></p><p><em>It is time to start the battle…</em><strong><em><br></em></strong><em>To end this era of sub-human chattel…</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=694027dde11" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Human Race]]></title>
            <link>https://medium.com/@tanyathakur6/the-human-race-75ed6df219d4?source=rss-e221ca4e11c4------2</link>
            <guid isPermaLink="false">https://medium.com/p/75ed6df219d4</guid>
            <category><![CDATA[violence]]></category>
            <category><![CDATA[devil]]></category>
            <category><![CDATA[freedom]]></category>
            <category><![CDATA[humanity]]></category>
            <category><![CDATA[hate]]></category>
            <dc:creator><![CDATA[Tanya Thakur]]></dc:creator>
            <pubDate>Sat, 19 Aug 2017 16:02:12 GMT</pubDate>
            <atom:updated>2017-08-19T18:11:06.900Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cunNjMqRFHIpodv_dfBH8g.png" /></figure><p><strong><em>A joke is criticised…<br>Violence is increasing its size…<br>But, making an intolerant remark is considered unwise…</em></strong></p><p><strong><em>We beat up dogs...<br>we vandalise trains...<br>Proving that we are insane...</em></strong></p><p><strong><em>Killings all around on something fake…<br>Religion above mankind is becoming everyone’s take…<br>And, the originator always manages to escape…</em></strong></p><p><strong><em>Treating destitute children like rag…<br>Debates over nuisance like wanting a state flag…<br>To the world we portray together we stand, but inside we hold north, south hatred tag…</em></strong></p><p><strong><em>Humanity is in pain…<br>Devils are at gain…<br>Effort of an angel is going in vain…</em></strong></p><p><strong><em>Where are we heading…<br>Towards freedom beheading???<br>Deep inside a pothole we are landing..</em></strong></p><p><strong><em>Making our agricultural land dry…<br>Igniting fire to hear people cry…<br>And at the end, voting for our nation to fly…</em></strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=75ed6df219d4" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>