<?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[QBox - Supercharge your chatbot’s intelligence - Medium]]></title>
        <description><![CDATA[More intelligent chatbots need better NLP training data - Medium]]></description>
        <link>https://medium.com/qbox-nlp-performance-tooling?source=rss----b525182b2ae1---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>QBox - Supercharge your chatbot’s intelligence - Medium</title>
            <link>https://medium.com/qbox-nlp-performance-tooling?source=rss----b525182b2ae1---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 06:41:54 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/qbox-nlp-performance-tooling" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Generate Chatbot training data with QBox — powered by Microsoft Turing NLG]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/generate-chatbot-training-data-with-qbox-powered-by-microsoft-turing-nlg-af69f7bfdd4d?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/af69f7bfdd4d</guid>
            <category><![CDATA[nlg]]></category>
            <category><![CDATA[turing]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[microsoft]]></category>
            <category><![CDATA[qbox]]></category>
            <dc:creator><![CDATA[Benoit Alvarez]]></dc:creator>
            <pubDate>Wed, 09 Jun 2021 16:22:20 GMT</pubDate>
            <atom:updated>2021-06-09T16:22:20.510Z</atom:updated>
            <content:encoded><![CDATA[<h3>Generate Chatbot training data with QBox — powered by Microsoft Turing NLG</h3><h3>Primary chatbot challenges</h3><p>One of the primary challenges when building any kind of chatbot is producing or obtaining high-quality, diversified training data. The training data that you use across your model’s intents will determine how readily your model picks up on a real user’s true intent when exposed to queries it’s never seen before. So no matter what chatbot framework you’re using (e.g. Microsoft LUIS, IBM Watson, etc.), having high-quality training data is a must. And it’s not just at the start of chatbot building that this makes a difference; for fine-tuning your model or re-evaluating your intent composition in more mature or larger chatbots, new and varied training data is always useful.</p><p>But coming up with strong training data is a common bottleneck for chatbot builders: it requires time and an understanding of how chatbots work. And sometimes chatbot builders get too close to the subject matter, and find it difficult to take a step back and think of more diverse training data.</p><p>This is where <a href="https://qbox.ai/?utm_source=Medium&amp;utm_medium=GenerateChatbotTrainingDataWithQBoxArticle&amp;utm_campaign=Microsoft%20Turing%20NLG">QBox</a>’s Suggest feature comes in. All a user has to do is ask <a href="https://www.youtube.com/watch?v=xGDXvk7QYmQ">QBox</a> for new utterances to be suggested to them across whichever intent(s) they want more data for. From these suggestions, the user can choose which, if any, to add to their model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*U2VPwfdP_sYwtN2I" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ybzd8aiLFpFvC9eqJ5sJlA.png" /></figure><h3>Turing and QBox secret sauce</h3><p>Behind the scenes, QBox scans your intents’ training data before integrating with Microsoft’s Turing transformer model. This 17-billion-parameter model has powerful natural-language generation competences that can produce novel utterances or training data. Within QBox, any novel utterances generated as suggestions to enhance an intent will preserve the integrity of that intent by staying within its semantic theme. Not only is the power of the Microsoft Turing model harnessed for these natural-language generations; QBox runs additional NLP processes to parse Turing’s output and present the user with only the most stable and useful utterance suggestions.</p><h3>Real example</h3><h4>Starting with what we have</h4><p>Let’s run through what this whole process looks like in QBox. Firstly, suppose you have an intent dealing with event accommodation and you’ve already come up with some training data/utterances for the intent. In the screenshot below, we can see on the left-hand side that QBox is not currently scoring this well: the primary QBox metrics of correctness, confidence, and clarity are all quite low.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*CSEvI3Q-5lVtS57e9BjlOg.png" /></figure><p>The QBox metrics are simple but very useful for understanding your chatbot’s performance. <em>Correctness</em> is the rate at which intents are being classified correctly. <em>Confidence</em> is how certain the model is that it has made the correct prediction. <em>Clarity</em> is the distinction between intents; lower scores mean your intents are more likely to be confused by novel queries because they’re too close to one another.</p><p>From this general overview, we can use the <strong>Experiment</strong> feature for the <strong><em>eventaccommodation </em></strong>intent. This feature has two functions; Suggest and Quick Analysis. If we switch over to the <strong>Suggest</strong> tab at the top of the screen and request suggestions, here’s what the initial output might look like.</p><p>All the utterances you see below are suggestions generated by Microsoft Turing NLG:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nlUnfJ-1uyhqPjIhQUGRig.png" /></figure><h4>Analyzing suggestions’ impact</h4><p>But wait — What if we haphazardly add the Turing-generated, QBox-parsed suggestions to our model and confuse it even further? Don’t worry: QBox has in-built functionality that handles this problem immediately. Quick Analysis allows you to test changes within QBox before pushing them to your NLP provider. Instead of fully committing to an updated model from the get-go, we can select the suggestions we’re interested in and run them through QBox’s Quick Analysis. Having a selective human in the loop between when Turing generates suggestions and when said suggestions are incorporated into the model is also important for filtering out potential bias and maintaining responsible use of AI. Let’s select just two suggestions that we’re interested in, as shown below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/653/1*2lhgtF5VHOgjCk7wCFuz4Q.png" /></figure><p>We can then perform Quick Analysis on the selected suggestions. This allows us to see what our scores (correctness, confidence, and clarity) would look like if we were to commit these suggestions to our model:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OHDINt68ShfDoBvv6HdDSg.png" /></figure><p>You can repeat these steps and continue using the Suggest and Quick Analysis features in tandem as much as you like. If you feel that the suggestions are too homogenous or you’d like more diversity in your training data, we have an <strong>I’m feeling adventurous</strong> checkbox in Suggest that will prompt greater variance from Turing’s natural-language generations.</p><h3>QBox and Microsoft Turing NLG in action</h3><p>Check out the video of <a href="https://qbox.ai/?utm_source=Medium&amp;utm_medium=GenerateChatbotTrainingDataWithQBoxArticle&amp;utm_campaign=Microsoft%20Turing%20NLG">QBox </a>and Microsoft Turing NLG in action</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FpbbJGKJT-KA%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DpbbJGKJT-KA&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FpbbJGKJT-KA%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/00b0ba2417ca72bb2d9e86834c0cde0d/href">https://medium.com/media/00b0ba2417ca72bb2d9e86834c0cde0d/href</a></iframe><h3>Conclusions</h3><p>The <a href="https://www.microsoft.com/en-us/research/blog/turing-nlg-a-17-billion-parameter-language-model-by-microsoft/">Microsoft Turing NLG model</a> allows for powerful, automated natural-language generation. Combined with our <a href="https://qbox.ai/?utm_source=Medium&amp;utm_medium=GenerateChatbotTrainingDataWithQBoxArticle&amp;utm_campaign=Microsoft%20Turing%20NLG">QBox </a>technology, it works very well within the context of chatbot building, which streamlines the automation process and provides meaningful metrics to measure its success.</p><p>This combination of automation and analysis provides great utility for users at all levels, from those at the early stages of chatbot building all the way to those working with the largest and most mature models — with ease of use remaining a fundamental priority.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=af69f7bfdd4d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/generate-chatbot-training-data-with-qbox-powered-by-microsoft-turing-nlg-af69f7bfdd4d">Generate Chatbot training data with QBox — powered by Microsoft Turing NLG</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[DevOps — QBox documentation]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/devops-qbox-documentation-40c6a14e7ae0?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/40c6a14e7ae0</guid>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[qbox]]></category>
            <category><![CDATA[mlops]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Benoit Alvarez]]></dc:creator>
            <pubDate>Fri, 29 May 2020 10:03:20 GMT</pubDate>
            <atom:updated>2021-10-14T16:00:02.917Z</atom:updated>
            <content:encoded><![CDATA[<h3>DevOps — QBox documentation</h3><p>This HTTP API allows you to run QBox tests and get results from them programmatically. All URLs are relative to the base URL, which is: <a href="https://api.qbox.ai/">https://api.qbox.ai</a></p><p>All endpoints return responses in JSON format and, when applicable, accept a request body in JSON format.</p><p><strong>This guide is an introduction to our API. For more comprehensive information, please check our </strong><a href="https://api.qbox.ai/openapi"><strong>Open API documentation</strong></a><strong>.</strong></p><h3>Authentication</h3><p>The API uses JWT token authentication. Unless otherwise specified all endpoints require authentication. An initial call must be made to get a token which can be used to authenticate subsequent requests.</p><p>To authenticate to the initial JWT token retrieval endpoint use the API username and API password found in the QBox user interface. To see this log in as an administrator account and navigate to Account and Settings -&gt; DevOps. If you don’t see the DevOps tab on the Account and Settings page then please contact your QBox sales representative. Credentials are set at the company level.</p><p>The token retrieved from the token retrieval endpoint is used as a bearer token, i.e. requests should include an “Authorization” header with a value of “Bearer [token]”. If the token expires then call the token generation endpoint again to get a new one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F1EYQn1F0URRKFsryjfJ-g.png" /></figure><h3>Endpoint</h3><h4>Authentication</h4><p>Retrieve authentication token</p><blockquote>URL: /authentication/authenticate</blockquote><blockquote>Type: POST</blockquote><p>This endpoint allows you to retrieve a JWT token to be used for authenticating to the other endpoints. It does not itself require a JWT token but uses the API username and API password from the QBox interface instead. See “Authentication” above for instructions on how to retrieve these.</p><p>Body:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/933/1*XD03QAsCVodV25peVnXHjw.png" /></figure><p>Response:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/933/1*MjXjALzHqwQ5I_wZAXOWlA.png" /></figure><h4>Retrieve current user</h4><blockquote>URL: /authentication/me</blockquote><blockquote>Type: GET</blockquote><p>This endpoint returns the name of the company the current authentication token is for. This can be used for validating an authentication token in a safe way.</p><p>Response:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/933/1*is-d9fdhaDa7ZM54MsK9rg.png" /></figure><h3>Tests</h3><h4>Run new test</h4><blockquote>URL: /test/run</blockquote><blockquote>Type: POST</blockquote><p>This endpoint allows you to start a new QBox test running. It returns an identifier which you can use to poll the test progress endpoint (below) to retrieve details of the test progress and test results.</p><p>Note that some setup may need to be done in the QBox front end for you to run certain types of test. If you are able to run a test in the QBox front end then you should be able to run it through the API.</p><p><a href="https://api.qbox.ai/openapi#/Test/post_Test_runfromurl"><strong>View API documentation</strong></a></p><p><strong>File types</strong></p><p>QBox will perform an HTTP GET request to the URLs you provide for the model file and cross-validation file. The body of the response to this request should be the content of the file with no extraneous content.</p><p>The file types are the same as those accepted by the QBox web interface. The model file can be one of the following:</p><ul><li>JSON file exported by IBM Watson Assistant or Microsoft LUIS</li><li>ZIP file exported by Google Dialogflow, Wit.ai or Amazon Lex</li><li>Markdown file (.md) generated by Rasa</li><li>TSV file generated by QnA maker</li><li>CSV file with the same format as the cross-validation file (see below)</li></ul><p>The cross-validation file may be a CSV file or a markdown file.</p><p>If it is a CSV file it should have the following format:</p><ul><li>First row should be a header row with values of “Question” and “Intent” (not necessarily in that order)</li><li>Each row should contain a single question/intent pair</li><li>Values can be enclosed in quotation marks and this should be done if a question or intent contains a comma</li></ul><p>If it is a markdown file (.md) it should follow the <a href="https://rasa.com/docs/rasa/nlu/training-data-format/">Rasa format specification</a>. Anything data in the file other than intents and their training samples will be ignored.</p><p>Response:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/931/1*Dx9qJukPoEqjdMuFTRcBJw.png" /></figure><p><em>Note that if your request fails validation then an HTTP 400 (Bad Request) response will be returned with the details of the problem. Correct requests return HTTP 200 (OK).</em></p><h3>Get test details</h3><blockquote>Url: /test/get?identifier=[identifier]</blockquote><blockquote>Type: GET</blockquote><p>Retrieves details of a test which was previously run through the API. If the test has finished then the results will be included as well.</p><p>The “identifier” querystring parameter should be set to the value of the “identifier” field in the response to the call to the “Run new test” endpoint.</p><p><a href="https://api.qbox.ai/openapi#/Status">View<strong> the result object.</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40c6a14e7ae0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/devops-qbox-documentation-40c6a14e7ae0">DevOps — QBox documentation</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[New model for rapid deployment of Covid-19-trained chatbots]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/new-model-for-rapid-deployment-of-covid-19-trained-chatbots-8806fb94a8ff?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/8806fb94a8ff</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[training-data]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[covid19]]></category>
            <dc:creator><![CDATA[Benoit Alvarez]]></dc:creator>
            <pubDate>Thu, 26 Mar 2020 17:04:40 GMT</pubDate>
            <atom:updated>2021-03-01T17:09:15.857Z</atom:updated>
            <content:encoded><![CDATA[<h3>Introduction</h3><p>The Covid-19-QBox-Chatbot Model is <strong>free</strong> pre-compiled chatbot model that addresses typical questions about the COVID-19 virus to help you handle high volumes of questions from your customers, partners and staff. You simply add it to your current chatbot training data.</p><p>We advise you to check that this new set of Intents, Entities &amp; Utterances do not conflict with your current model. To check this out for free, visit <a href="https://qbox.ai/?utm_source=Medium&amp;utm_medium=Covid19%20Bot%20Article&amp;utm_campaign=Learn%20more">Qbox.ai</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Mw9xMMU9V5MyG49aol7SIA.png" /></figure><h3>Download</h3><p>Available for:</p><ul><li><a href="https://github.com/benoitalvarez/Covid-19-QBox-ChatbotModel/tree/master/Models/Microsoft-LUIS">Microsoft LUIS</a></li><li><a href="https://github.com/benoitalvarez/Covid-19-QBox-ChatbotModel/tree/master/Models/Microsoft-QnAmaker">Microsoft QnA Maker</a></li><li><a href="https://github.com/benoitalvarez/Covid-19-QBox-ChatbotModel/tree/master/Models/Google-Dialogflow">Google Dialogflow</a></li><li><a href="https://github.com/benoitalvarez/Covid-19-QBox-ChatbotModel/tree/master/Models/IBM-Watson-Assistant">IBM Watson Assistant</a></li><li><a href="https://github.com/benoitalvarez/Covid-19-QBox-ChatbotModel/tree/master/Models/Rasa">RASA</a></li></ul><p>With this supplementary model, your chatbot service will help alleviate pressure on your front line staff by filtering many of the common queries that are being received day each day as this crisis escalates. To help we’ve built a chatbot focussed on queries related to the COVID-19 outbreak and its likely impact on a business.</p><h3>How the model is structured</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/632/1*Ay8v_etdRwAMDbqXAN0YTQ.png" /></figure><p>The above table shows the full list of “intents” (answers) to queries that the model has been trained to address. It primarily covers general FAQ-type queries to make it more widely applicable, but also includes several business-related queries. The model avoids more sensitive topics such as customer complaints or addressing mental health issues during self-isolation, since those issues should be handled by a trained professional, rather than a chatbot.</p><p>Please note, this model is only for returning intents — the content of the answers themselves should be managed elsewhere. We did not want to include these type of answers, as the situation is still fluid, with Government and health advice is varying by day and country. Businesses can provide their own tailored answers to the “company_” intents as well.</p><p>Also included is a “small-talk” intent for the purpose of making it clear this is a chatbot that offers the user the chance to speak with a human. It’s important to be honest about the nature of your chatbot, and to offer speaking with a trained professional as an alternative. There are many API services and other open source-models, should you wish to expand the small-talk intents further, but I opted to keep this limited due to the serious nature of the subject matter.</p><p>Entities, although not available in all NLP engines, can be useful for both addressing synonyms more easily (a user might say “what is Corona?” or “what is COVID-19?”, and for creating answer variants. You can for example, create a specific answer for faq_vulnerable where a pregnancy entity is also detected — e.g. “I’m pregnant, am I at risk?”</p><h3>Github link</h3><p><a href="https://github.com/benoitalvarez/Covid-19-QBox-ChatbotModel">benoitalvarez/Covid-19-QBox-ChatbotModel</a></p><p>This is free to use. Enjoy and stay safe</p><p><a href="https://qbox.ai/?utm_source=Medium&amp;utm_medium=Covid19%20Bot%20Article&amp;utm_campaign=Learn%20more">QBox </a>and <a href="https://volume.ai/">Volume </a>teams</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8806fb94a8ff" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/new-model-for-rapid-deployment-of-covid-19-trained-chatbots-8806fb94a8ff">New model for rapid deployment of Covid-19-trained chatbots</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[QBox September 2019 Update]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/qbox-september-2019-update-686ebdc70f3c?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/686ebdc70f3c</guid>
            <category><![CDATA[dialogflow]]></category>
            <category><![CDATA[microsoft-luis]]></category>
            <category><![CDATA[qbox]]></category>
            <category><![CDATA[npl]]></category>
            <category><![CDATA[ibm-watson]]></category>
            <dc:creator><![CDATA[Benoit Alvarez]]></dc:creator>
            <pubDate>Mon, 16 Sep 2019 09:32:03 GMT</pubDate>
            <atom:updated>2019-09-16T09:35:38.578Z</atom:updated>
            <content:encoded><![CDATA[<h3>QBox user base continues to grow by over 20% MoM</h3><p>During August, we had 155 users run a total of 432 tests across 120 projects. Combined, they saw impressive improvements across their <a href="http://www.qbox.ai/">QBox </a>scores:</p><blockquote>Correctness: 480 points</blockquote><blockquote>Confidence: 607 points</blockquote><blockquote>Clarity: 727 points</blockquote><figure><img alt="Chatbot Provider Market Share" src="https://cdn-images-1.medium.com/max/852/1*3Ld1PMmlT7e1U_JB7Qql_w.png" /><figcaption>Chatbot Provider Market Share</figcaption></figure><p>We welcomed new customers including:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*8VmeStpVkghI5Y9ci30HRQ.png" /></figure><h3>Released in early September (QBox 2.3)</h3><h4>Improvements on new features within IBM Watson Assistant, Microsoft LUIS &amp; Dialogflow from Google</h4><p>As these NLP providers rapidly develop their functionalities and features, we continuously ensure <a href="http://www.qbox.ai/">QBox </a>supports them.</p><p><em>IBM Watson Assistant</em></p><ul><li>Support for pattern type entities</li></ul><p><em>Microsoft LUIS</em></p><ul><li>Support for regex type entities</li><li>Support for patterns</li><li>Support for entity roles</li><li>Enhanced support for LUIS prebuilt domain entities and intents</li></ul><h4>Also</h4><p>Warning on the job summary page are now displayed as icons rather than as a separate tab</p><h3>Coming up end of September (QBox 2.3)</h3><ul><li>Exclude intent from analysis (i.e. button intent)</li><li>Testing Google Dialgflow mutli-culture model separately</li><li>Moving our new Beta Explain functionality to all Enterprise accounts</li><li>Add a collaboration functionality at intent level. Being able to add comments and notes</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=686ebdc70f3c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/qbox-september-2019-update-686ebdc70f3c">QBox September 2019 Update</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[QBox version 2.5 released note]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/qbox-version-2-5-released-note-4ccf114a4752?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/4ccf114a4752</guid>
            <category><![CDATA[qbox]]></category>
            <category><![CDATA[amazon-lex]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Benoit Alvarez]]></dc:creator>
            <pubDate>Thu, 20 Dec 2018 16:22:18 GMT</pubDate>
            <atom:updated>2019-01-03T14:51:14.353Z</atom:updated>
            <content:encoded><![CDATA[<h3>QBox version 2.5 release notes</h3><p>This latest release is packed with new and improved features.</p><ul><li>Amazon Lex alpha release</li><li>Facebook WIT and Google Dialogflow test speed improvement</li><li>Better insight of our simple view</li><li>Pricing change plus PayPal support</li><li>Time estimation and notification</li><li>Monitoring time saved and IBM Watson dev-ops</li><li>Help page, tip of the day</li><li>Payment in Euro and US Dollars</li></ul><h3>Welcome to Amazon LEX</h3><p>After a successful beta test, we are pleased to confirm that Amazon Lex is now fully supported. Anyone with a Lex model can now analyse its performance and improve their customer experience using QBox.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/320/1*mpoQlfyNAkuiRENYpXQjwQ.png" /></figure><h3>Time is money</h3><p>We know the faster you get your trained data test results, the faster you can modify your NLP Data Model and improve accuracy. We are continually improving test speeds across four of the NLP vendors.</p><ul><li>Amazon Lex and Dialogflow are now 2x faster to run.</li><li>Wit.ai is now 3x times to run.</li><li>Rasa.NLU is now 5x faster to run.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/914/1*cRUSxjm6YIlCt0wb5GMSuQ.png" /></figure><h3>Redesign for better understanding</h3><p>The intent details page now shows a straightforward visualisation of individual test results that are causing your intent to underperform. The visualisation shows where your intent was in the NLP confidence ranking, and where QBox thinks it should be. As before, you can drill down to find out more</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*h4TKZphrzgyZblOS7O9uGQ.png" /></figure><h3>Pricing change, plus PayPal support</h3><p>After may requests from customers we now support PayPal. We also clarified each tier, and the functionalities you can have access to. Finally we added the possibility to payment in Euro and US Dollars for our clients.</p><p>See the table. For more information, visit <a href="https://qbox.ai/Home/Pricing">https://qbox.ai/Home/Pricing</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IgHq5GNuheTswDuDCSzpnQ.png" /></figure><h3>Monitoring time saved and IBM Watson dev-ops</h3><p>The monitoring page now tells you explicitly how much time QBox has saved by automatically reviewing your chatbot’s user interactions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n0Eg_H8al4IJQv2h_iAPTg.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/982/0*gfF2qMiqxsgwxpnJ.png" /></figure><p>When reviewing and improving your training data for IBM Watson Assistant, your only option was to modify your live model or to create a clone of your model manually. QBox now manages it all. Now you simply review your live model, identify user interactions and where and when to improve your model, QBox clones the model, enabling you to modify it until it reaches the performance levels you expect. Once ready, QBox places your cloned model live automatically in your production chatbot, resulting in zero downtime.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ccf114a4752" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/qbox-version-2-5-released-note-4ccf114a4752">QBox version 2.5 released note</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[QBox version 2.4 released note]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/qbox-version-2-4-released-note-58960e63fb12?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/58960e63fb12</guid>
            <category><![CDATA[qbox]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[microsoft-luis]]></category>
            <dc:creator><![CDATA[Benoit Alvarez]]></dc:creator>
            <pubDate>Mon, 15 Oct 2018 11:26:33 GMT</pubDate>
            <atom:updated>2019-01-03T14:51:01.923Z</atom:updated>
            <content:encoded><![CDATA[<h3>QBox version 2.4 release notes</h3><p>The biggest improvement in this release is the introduction of support for Microsoft LUIS in QBox Enterprise (QE). This works in exactly the same way as IBM Watson™ Assistant integration. The performance of both providers is very similar.</p><h3>Other improvements in this sprint</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rpQoMlHOxwmcvj01JTrQeQ.jpeg" /></figure><p>We’ve also added support for the new IAM (Identity and Access Management) authentication mechanism in IBM Watson Assistant. IBM is migrating to this authentication mechanism on a region-by-region basis (the UK has not been migrated yet), and service instances, which were created prior to the migration, will not have their authentication credentials changed.</p><p>API connections to your different NLP providers are now stored on the ‘company’ level instead of the ‘user’ level.</p><p>The QBox Enterprise (QE) review list has been redesigned to make the functionality clearer and to reduce the number of statistics displayed to show clearly how QE helps reduce the workload when reviewing live logs from user interactions with its “Auto-marked” algorithm.</p><p>QE now includes questions in the review list, where the model returned the correct answer but where the confidence was below the set threshold. You can disable this and can change the confidence threshold for your model (by default it is set to 40%). Questions that return the right intent but with low confidence are also highlighted in the review list.</p><p>The QE list can be reviewed by multiple users at the same time to increase efficiency for large enterprises.</p><p>The “Score” tab has been renamed to “Intents” on the job summary page.</p><h3>QBox vs QBox Enterprise</h3><p>Below is a summary of QBox and QBox Enterprise</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nqI0eCxKA71PZNSNEjccHw.png" /></figure><p>QBox helps to measure how well your chatbot understands your user questions. You will be able to refine your training data in your model and measure its performance.</p><p>QBox Enterprise (QE) is utilised post chatbot go live, where you’ll need to audit or review a batch of interactions. QE will automate the majority of the review, and will optimise the list so that the data model analyst or SME can review interactions that are the most likely to be incorrect.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Lcao3NPHQdTLyRj6w_zRkA.png" /></figure><blockquote>To try for free those new improvement or to book a demo, please visit <a href="https://qbox.ai/">https://qbox.ai</a> now.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=58960e63fb12" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/qbox-version-2-4-released-note-58960e63fb12">QBox version 2.4 released note</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Improving your Microsoft LUIS model]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/improving-your-microsoft-luis-model-fbc0cc180625?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/fbc0cc180625</guid>
            <category><![CDATA[luis]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[microsoft-luis]]></category>
            <category><![CDATA[intent]]></category>
            <dc:creator><![CDATA[QBox - NLP performance tooling]]></dc:creator>
            <pubDate>Fri, 14 Sep 2018 09:36:32 GMT</pubDate>
            <atom:updated>2018-09-14T09:47:29.638Z</atom:updated>
            <content:encoded><![CDATA[<iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F9qTTor-NbIM%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D9qTTor-NbIM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F9qTTor-NbIM%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/829ed31a1265b0fcf6af70f7cf6919eb/href">https://medium.com/media/829ed31a1265b0fcf6af70f7cf6919eb/href</a></iframe><p>This video provides an overview on how and where to improve the performance of your chatbot platform using LUIS through <a href="http://www.qbox.ai/">QBox</a>. Although the video is 35 minutes long, by the end of it you will have learnt how to:</p><ol><li><strong>Fix poorly performing intents</strong></li><li><strong>Identify potential future problems with training data within intents, and fixing before they become an issue further down the line</strong></li><li><strong>Recognise and deal with model regression</strong></li></ol><blockquote>The step by step guide is ideal for first time users of <a href="http://www.qbox.ai/">QBox</a> or those who have just tested their model through <a href="http://www.qbox.ai/">QBox</a> and are <strong>daunted </strong>by the initial test results and unsure where to start.</blockquote><p>The model used in the demo, built in Microsoft LUIS, is a small model based on banking questions and has 18 intents. The initial <a href="http://www.qbox.ai/">QBox </a>test results show many of the intents are not performing very well. We guide you through where to start, and explain why <a href="http://www.qbox.ai/">QBox </a>has highlighted problem intents, and how they can be improved upon by adding or amending training questions. We also explain any model regression that may be caused through the additions, and the issues with false positives and how these can be fixed.</p><p>The video demonstrates that by focusing on the worst performing intent, taking small steps to fix it, testing again in <a href="http://www.qbox.ai/">QBox</a>, addressing any model regression and then further testing, you can gradually improve overall accuracy and confidence. This will then make your model more stable and effective.</p><p>Let us know any comments.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/445/1*zQP-LFmhpBhl-p9XjaFQ8A.gif" /></figure><p>Thanks</p><p>QBox team.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fbc0cc180625" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/improving-your-microsoft-luis-model-fbc0cc180625">Improving your Microsoft LUIS model</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[QBox Enterprise — model life cycle monitoring]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/nlpqbox-enterprise-model-life-cycle-monitoring-87a95e32877?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/87a95e32877</guid>
            <category><![CDATA[nlp-data-model]]></category>
            <category><![CDATA[watson-assistant]]></category>
            <category><![CDATA[training-data]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[watson-conversation]]></category>
            <dc:creator><![CDATA[QBox - NLP performance tooling]]></dc:creator>
            <pubDate>Fri, 07 Sep 2018 09:31:24 GMT</pubDate>
            <atom:updated>2018-09-07T13:56:12.995Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*igBJJbpNOFchZxKEx7o9jw.png" /></figure><p><em>Conversational platforms built to handle thousands of ‘chats’ daily will create a challenge for organisations trying to sift through a growing volume of data necessary to validate the accuracy, and refine where necessary, the performance of the NLP data model.</em></p><p><em>QBox Enterprise uses intelligent algorithms and automates the process of tagging and ranking the system’s responses, identifying those that have likely not performed well.</em></p><p><em>This dramatically reduces the complexity and time required by data scientists and NLP data model analysts to decipher where the training priorities lie and pinpoint areas of immediate or urgent attention.</em></p><p><em>QBox Enterprise is the latest enhancement to the QBox Chatbot testing solution that helps to optimise performance, efficiency and ensure the best experience for the customer.</em></p><h3>Model life cycle monitoring, QBox missing piece</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*igoqQZLWIHQBAW3ScxWYcg.png" /></figure><p>Currently, it can be arduous monitoring the NLP data model to verify whether it classifies correctly customer queries. Chatbot and conversational platform developers have little or no options for managing the life cycle of their model from training and performance tuning, to holistic monitoring. QBox’s first release included a core training and performance tuning tool. In this new release, QBox Enterprise, intelligent monitoring is added to Training &amp; Testing-as-a-Service for NLP data models.</p><p>Chatbot system administrators receiving high volumes of customer queries per day, will soon have to address the challenge of monitoring and assessing the model’s performance in order to maintain quality of service and a means to continually improve.</p><h3>New release</h3><p>QBox Enterprise includes a new user interaction monitoring module. This release includes:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*R-g7gJMlesEk4IAQcHfqJw.png" /></figure><ul><li><strong>Monitoring live models: </strong>NLP data model analysts or system administrators can set QBox to monitor live customer interactions from their NLP service provider</li><li><strong>Intelligent self-learning scoring algorithm from each interaction: </strong>Once customer interactions have been collated, QBox’s intelligent self-learning algorithm scores and automatically classifies interactions. As a result, interactions will be marked as ‘correct’ (so no need to be reviewed), ‘likely correct’, ‘likely incorrect’ and ‘incorrect’ (so needs to be reviewed as a priority).</li><li><strong>Automatic sampling for Data Model Managers using QBox’s proprietary algorithm: </strong>The sample module uses ‘scoring’ and classification to select an optimal list of unbiased customer interactions that are flagged for the Data Model Manager to review.</li><li><strong>Automatic processing to measure, compare and help fix wrong predictions:</strong> Once the Data Model Manager has marked the proposed review list, QBox isolates the interaction marked ‘incorrect’ and runs a benchmark QBox test. QBox alerts the Data Model Manager as to where they need to fix the model. QBox will then compare the corrected model against the incorrect marked interaction, whilst also checking for any other instances and occurrences of global regression.</li></ul><p>If you want to test this new feature, please contact us at <a href="https://qbox.ai/">https://qbox.ai/</a></p><p>Note: This new feature release is only available for IBM Watson™ Assistant. Other NLP service providers will be announced soon.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=87a95e32877" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/nlpqbox-enterprise-model-life-cycle-monitoring-87a95e32877">QBox Enterprise — model life cycle monitoring</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[QBox version 2.2 released note]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/qbox-version-2-2-released-note-814850e40b3e?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/814850e40b3e</guid>
            <category><![CDATA[microsft-luis]]></category>
            <category><![CDATA[ibm-watson-conversation]]></category>
            <category><![CDATA[ibm-watson-assistant]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[rasa-nlu]]></category>
            <dc:creator><![CDATA[QBox - NLP performance tooling]]></dc:creator>
            <pubDate>Thu, 09 Aug 2018 15:12:45 GMT</pubDate>
            <atom:updated>2018-09-03T14:05:38.359Z</atom:updated>
            <content:encoded><![CDATA[<p>We have just finished on our 2.2 release.</p><h3><strong>SECRET…</strong></h3><p>The main feature of this <a href="https://qbox.ai/">QBox </a>sprint is the first part of our secret new feature that will be released in second half of September. Watch this space</p><p>Other functionality we can disclose which has been deployed in this release:</p><h3><strong>RASA NLU</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/322/1*W9IJaKpTKhj9Tp9o6Sc6zw.png" /></figure><p>The Rasa architecture has been restructured to separate the code which runs Rasa jobs from the main server. This means that the site doesn’t slow down when you run Rasa jobs. It does make Rasa jobs a bit faster as well. Models with under 1000 questions should run in less than about 20 minutes, however the massive 4,000 question models which have been run against Rasa recently will still take on the order of 90 minutes to run. So still so way to go to equal the speed of Watson Assistant or Microsoft Luis.</p><h3><strong>WATSON</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/329/1*yRtPAGWhIRW0Quz-SQ-pKQ.png" /></figure><p>We now have partial support for the new beta contextual entities in Watson Assistant. If you upload a Watson model with contextual entities in it and run it against Watson then the model will be trained with those contextual entities present. If you upload a Dialogflow or Rasa model which has entity values marked in training data and run against Watson then the entity values in training data will be converted, and vice versa. Read <a href="https://medium.com/@desmarchris">Chris Desmarais</a> formore information on this Watson functionality <a href="https://medium.com/ibm-watson/watson-assistant-new-capabilities-c40aed97e5fd">https://medium.com/ibm-watson/watson-assistant-new-capabilities-c40aed97e5fd</a></p><p>This does not work for Luis or Wit.ai models — these will not be converted to or from Watson. If you upload a Watson model and run it against Dialogflow then it will look for occurrences of entities in training data and mark them, in addition to the new contextual entities. It will not do this for any other provider conversion.</p><h3><strong>NLP PROVIDER COMPARISON</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o3frJ-PwZ28c9d0kIXAiRQ.png" /></figure><p>The display of deltas for multiprovider jobs on the job list page and the comparison popup has been changed. Deltas on the comparison popup are now differences from the best provider rather than from the previous job run.</p><h3><strong>OTHER</strong></h3><ul><li>The privacy policy has been updated.</li><li>A few bugs have been fixed, including one which affected a live user yesterday.</li><li>The job summary page should be a little faster to load.</li></ul><p>Try QBox for free at <a href="https://qbox.ai/">https://qbox.ai</a></p><p>Or contact us at info@qbox.ai</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=814850e40b3e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/qbox-version-2-2-released-note-814850e40b3e">QBox version 2.2 released note</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Introduction of QBox — January 18 update]]></title>
            <link>https://medium.com/qbox-nlp-performance-tooling/introduction-of-qbox-january-18-update-61af94e250c6?source=rss----b525182b2ae1---4</link>
            <guid isPermaLink="false">https://medium.com/p/61af94e250c6</guid>
            <category><![CDATA[ibm-watson-conversation]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[ibm-watson-assistant]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[microsoft-luis]]></category>
            <dc:creator><![CDATA[QBox - NLP performance tooling]]></dc:creator>
            <pubDate>Wed, 08 Aug 2018 16:47:13 GMT</pubDate>
            <atom:updated>2018-07-17T11:29:26.950Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/900/1*7T2ObfkLNXhTEXsNqa3ucA.png" /></figure><p>Volume is a Gartner-recognised developer of conversational platforms with a number of live instances in market. The company’s early work in this field quickly revealed that the main NLP providers’ natural-language data models were essentially black boxes, and that tools to help assess performance and the impact of scaling simply weren’t available.</p><p>The key NLP providers give access to simple tooling to create data models but none enable developers to measure the performance and quality of the model. When trying to scale and train their models, developers are faced with manual, high-capital-cost, error-prone and slow process. With customer experience the key measurement of success, many brands don’t yet have the confidence or capability to deploy chatbots more widely.</p><p>Our own experience of working with natural-language data models led us to develop QBOX: a middleware technology-agnostic corpus-management tool that allows developers to quantify the quality of natural-language models and the impact of training and scaling.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WZ8WluApXidOachpYtUmeQ.jpeg" /></figure><p><strong>QBOX ADDRESSES KEY CHALLENGES WHEN<br>WORKING WITH NL DATA MODELS.</strong></p><p><strong>MEASURABILITY: </strong>Benchmark your data model using automatically generated KPIs to validate performance.</p><p><strong>VISUALISATION: </strong>Easily interpret and analyse your test results with different views to facilitate interpreting and analysing your test results.</p><p><strong>IDENTIFICATION: </strong>Improve the accuracy of your natural-language data model by detecting poor-performing intents.</p><p><strong>SCALABILITY: </strong>Rapidly scale your data model and maintain performance when new intents are added.</p><p><strong>AFFORDABILITY: </strong>Significantly reduce your costs by simplifying maintenance processes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*b8X2uA9JBXHmhOWI1Q8l8g.jpeg" /></figure><p><strong>WHAT THE QBOX TEAM HAS BEEN UP TO:</strong></p><ul><li>We continue to improve QBOX based on Gartner analysts’ and external users’ feedback.</li><li>The QBOX go-to-market strategy is to offer a vendor-agnostic solution. Currently, QBOX supports IBM Watson and Microsoft LUIS services.</li><li>Dialogflow is currently in alpha, transitioning to beta by the end of Q1 2018.</li><li>WIT.ai alpha is planned for mid-January 2018.</li></ul><p><strong>ADDITIONAL FUNCTIONALITIES:</strong></p><ul><li><strong>Three-dimensional scoring:</strong> The new scoring method evaluates the quality of the model and intents from three independent viewpoints: correctness, confidence and prediction stability. This equates to how correctly we can predict the right intent, how confident we are about this prediction, and finally, how stable over time this prediction will be.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/365/1*RILAUc4AS6NvUELAuA6okw.png" /></figure><ul><li><strong>Concept recognition and mapping:</strong> This feature extracts concepts from the training data of an intent, to visualise whether a particular concept requires reinforcement.</li><li><strong>Payment:</strong> Payment functionality and tiered access are now available.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=61af94e250c6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/qbox-nlp-performance-tooling/introduction-of-qbox-january-18-update-61af94e250c6">Introduction of QBox — January 18 update</a> was originally published in <a href="https://medium.com/qbox-nlp-performance-tooling">QBox - Supercharge your chatbot’s intelligence</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>