<?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 Dan Sandu on Medium]]></title>
        <description><![CDATA[Stories by Dan Sandu on Medium]]></description>
        <link>https://medium.com/@dan_16362?source=rss-4aa744e1a5f0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*JYvDyYExNbgw9QBz.jpg</url>
            <title>Stories by Dan Sandu on Medium</title>
            <link>https://medium.com/@dan_16362?source=rss-4aa744e1a5f0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 11:48:02 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dan_16362/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[A quick guide for effective chatbot training in customer service]]></title>
            <link>https://blog.chatbotslife.com/a-quick-guide-for-effective-chatbot-training-in-customer-service-ad75ed768390?source=rss-4aa744e1a5f0------2</link>
            <guid isPermaLink="false">https://medium.com/p/ad75ed768390</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[chatbot-development]]></category>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[bots]]></category>
            <dc:creator><![CDATA[Dan Sandu]]></dc:creator>
            <pubDate>Mon, 04 Mar 2019 14:34:01 GMT</pubDate>
            <atom:updated>2019-03-04T14:34:01.832Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*y2bPRztL7A7mhnfgzZqw7Q.png" /></figure><p>The quality of a <a href="https://chatbotslife.com/">chatbot</a> depends on a large number of factors from the design of the conversation flow to how well the responses are crafted or how easy is for the customer to speak to a human.</p><p>Among all these factors, there is one thing that any customer support <a href="https://chatbotslife.com/">chatbot</a> needs to do very well from the beginning and keep getting better at it: understand your customers questions. Due to the nature of human language, training a chatbot effectively to correctly map phrases and sentences to what a customer actually means (his intent) is particularly challenging and one of the main reasons chatbots fail.</p><figure><a href="https://www.eventbrite.com/e/chatbot-ai-conference-tickets-55380774387"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ac4Wm94IFfL37KAAXnEUXQ.png" /></a></figure><h3>Getting the training data</h3><p>Once you have the conversation flow laid out and you know which customer questions need to be answered, you’ll start training your chatbot for each of these questions or intents. In order to do this you need to gather a comprehensive dataset of customer utterances or examples of sentences used by customers to ask a question or mention a problem. There are several ways to get this data and an obvious one to get started is to have the <a href="https://chatbotslife.com/">chatbot</a> developer come up with what he or she thinks the customer would say. This approach obviously is very limited and in most cases only suitable for very basic use cases.</p><h3>Top 4 Most Popular Bot Design Articles:</h3><blockquote><a href="https://chatbotslife.com/why-conversational-context-is-the-most-powerful-tool-you-can-give-your-chatbot-19d9d8656697">1. Why Conversational Context is the Most Powerful Tool you can give your Chatbot</a></blockquote><blockquote><a href="https://chatbotslife.com/emotionally-intelligent-chatbots-part-1-getting-emotions-from-users-face-4c07447dada8">2. Emotionally Intelligent ChatBots — Part 1: Getting emotions from user’s face</a></blockquote><blockquote><a href="https://chatbotslife.com/emotionally-intelligent-chatbots-part-2-getting-emotions-from-users-voice-3b29048792b7">3. Emotionally Intelligent ChatBots — Part 2: Getting emotions from user’s voice</a></blockquote><blockquote><a href="https://chatbotslife.com/chatbot-conference-2019-in-nyc-8b010058f8f9">4. Chatbot Conference 2019 in NYC</a></blockquote><p>An alternative way would be to have a limited <strong>cohort of actual users</strong> (employees, interns, SMEs) interact with an alpha version of the chatbot. The data generated by these interactions can be used for further training and improving chatbot understanding. While this approach would produce better quality and more varied data, it is very expensive, takes a lot of time and is hard to scale.</p><p><strong>Publicly available datasets </strong>can be another source of training data for your chatbot. Today there is a significant number of these datasets available and a simple google search can retrieve some of them. I’m listing a few here for reference:</p><ul><li><a href="https://data.stackexchange.com/">Stackoverflow</a>: 7.3 million stackoverflow questions</li><li><a href="https://www.kaggle.com/datasets?sortBy=hottest&amp;group=public&amp;page=1&amp;pageSize=20&amp;size=all&amp;filetype=all&amp;license=all&amp;tagids=11208">Kaggle Linguistic Datasets</a>: over 300 sources in multiple languages</li><li><a href="https://snap.stanford.edu/data/web-Amazon.html">Amazon Reviews</a>: Stanford collection of 35 million amazon reviews</li><li><a href="http://webscope.sandbox.yahoo.com/catalog.php?datatype=l">Yahoo! Questions and Answers</a>: Yahoo! Answers corpus with 4.4 millions questions and their answers</li></ul><p>This data is obviously rich and varied but on the other hand very difficult to navigate and extract what is relevant for your chatbot use cases. It depends on your business domain, but excepting few basic ones (like changing a password for example) most of the questions and problems your customers need help with won’t be found in publicly available data.</p><p>There is another source of data, both rich and relevant, hidden in the support helpdesk or chat system. The <strong>past conversations with your customers</strong> is a real treasure for chatbot development. In there you can find all the questions and problems that your customers struggle and need help with. The data includes but is not limited to: support tickets, transcripts of live chats, emails and agent notes. Using this data though requires <a href="https://www.akodot.com/">exporting it from your helpdesk or chat vendor</a> and transforming it into a format that is usable for chatbot training. An upfront and manual effort that is not negligible.</p><h3>Structuring your training data</h3><p>For each customer question to be handled by your chatbot you need a rich set of examples of sentences the customers might use. Therefore the next job after securing the sources of data is to extract these examples and manually <a href="https://www.akodot.com/">annotate</a> them with the relevant intent.</p><p>If you have tens of thousands of past conversations to choose from, this task can rapidly become daunting, like looking for a needle in a haystack. You need a quick and efficient way to cut through the noise. An approach that often works is coming up with a set of words that you know the customers use when asking a question.</p><blockquote>If you have tens of thousands of past conversations to choose from, annotating them can rapidly become daunting, like looking for a needle in a haystack.</blockquote><p>Take for example the chatbot use case of helping a customer changing his password. A quick search for the word “password” will surface a reduced set of examples of sentences where you can pick from and <a href="https://www.akodot.com/">annotate</a> with this intent. Obviously customers might use other words than “password”, so you’ll need to try a number of synonyms or words used in the same context (like “pwd”, “access”, “username” etc) in order to get a high variety of examples for this intent.</p><h3>Train the chatbot</h3><p>Once you have the data structured in a format usable for chatbot training, you can start the actual training now. The workflow depends on what chatbot development platform you’re using but generally it includes creating the intents for each use case and <a href="https://www.akodot.com/">giving examples of sentences</a> for each intent. If the conversation flow for an intent has a few levels you’ll have to repeat this step for each level, or each time you’re expecting a response from the user.</p><p>This step is relatively straightforward but if you have many intents to support and a large number of training examples for each intent, it can become a heavy task. Some platforms support uploading csv or text files with example sentences. Be careful though with moving and storing files, in case they might contain sensitive information.</p><h3>Deploy, measure and improve</h3><p>The process of getting the chatbot in production depends on the platform and channel(s) used, and is out of the scope of this guide. However, there is one thing to keep in mind: how well your chatbot understands your customers will be limited at the beginning. You will have to put in place a feedback loop where sentences that couldn’t be correctly classified to an intent should be extracted, annotated and fed back to the chatbot for training.</p><p>Additionally you should continuously train the chatbot with data from real customers conversations, or any other relevant source.</p><h3>A chatbot is as good as its training</h3><p>Thanks to recent advances in Natural Language Understanding made available by major development platforms such as Google, Microsoft or IBM, building a customer service chatbot has become accessible to many businesses who are looking to improve and scale the way they help their customers. However, no matter what the platform is, there is one common task that needs to be done repetitively: training your chatbot with the best quality data you can get.</p><blockquote>there is one common task that needs to be done: training your chatbot with the best quality data you can get</blockquote><p>The process of training a chatbot requires a large amount of manual effort and can often feel overwhelming. In order to speed it up you need an <a href="https://www.akodot.com/">efficient way</a> to collect training data from past customer conversations, annotate it rapidly for each customer question and upload it to the chatbot development platform of choice.</p><p><em>Your chatbot is as good as its training, and the quality of the training is only as good as the training data is. Our mission at </em><a href="https://www.akodot.com/"><em>Akodot</em></a><em> is to significantly improve the way chatbots are being trained using data from real customer conversations.</em></p><h3>Don’t forget to give us your 👏 !</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/780/0*kNtFwY-TOd4tOZG7" /></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F3e59ef%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F3e59ef&amp;image=http%3A%2F%2Fapi.screenshotlayer.com%2Fapi%2Fcapture%3Faccess_key%3Dfe59908dad3baab69ffab249a2224b03%26viewport%3D1024x612%26width%3D1000%26url%3Dhttps%253A%252F%252Fupscri.be%252F3e59ef%253Fscreenshot&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/7078d8ad19192c4c53d3bf199468e4ab/href">https://medium.com/media/7078d8ad19192c4c53d3bf199468e4ab/href</a></iframe><figure><a href="https://chatbotslife.com/bot-communities-mastermind-group-d2dae9876709#.53x0py6ou"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*6XUspT9JOSq0w0Fi35HIaA.png" /></a></figure><figure><a href="https://m.me/ChatbotsLife"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*c1LDMH5vbnIz9rmAka8Hwg.png" /></a></figure><figure><a href="https://chatbotslife.com/chatbot-development-hire-top-ai-chatbot-developers-c8b45ef7f207"><img alt="" src="https://cdn-images-1.medium.com/max/255/1*D0Jf3dI6ZThtqcfwDYY7mg.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ad75ed768390" width="1" height="1" alt=""><hr><p><a href="https://blog.chatbotslife.com/a-quick-guide-for-effective-chatbot-training-in-customer-service-ad75ed768390">A quick guide for effective chatbot training in customer service</a> was originally published in <a href="https://blog.chatbotslife.com/">Chatbots Life</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>