- 5/7/2025
Welcome to this hands-on lab session on Natural Language Processing (NLP) using Azure AI Services! Whether you’re preparing for the AI-900 Microsoft Certification or exploring how NLP works, this step-by-step demonstration will help you understand how to harness Azure AI to process and analyze human language data.
🔍 What You'll Learn in This Video:
1️⃣ Introduction to NLP and its significance in the AI landscape.
2️⃣ Azure Language Services Overview: Features and tools for NLP tasks.
3️⃣ Step-by-step guide to:
Text analysis (sentiment analysis, key phrase extraction, etc.).
Named Entity Recognition (NER).
Language detection and translation.
Conversational AI capabilities.
4️⃣ Real-world applications of NLP in industries like customer service, healthcare, and e-commerce.
🛠️ Who Is This For?
Beginners curious about Natural Language Processing.
Professionals preparing for the AI-900 exam.
AI enthusiasts and developers looking to integrate language intelligence into their applications.
📌 Key Features Highlighted:
✅ Build intelligent apps using Azure AI NLP tools.
✅ Real-time text analysis and language understanding.
✅ Simple, no-code/low-code solutions in the Azure portal.
✅ Scalable NLP services for enterprise-grade projects.
💡 Take your understanding of NLP to the next level with Azure AI!
Explore Our Other Azure Courses and Practice material on: https://www.youtube.com/@skilltechclub
🔍 What You'll Learn in This Video:
1️⃣ Introduction to NLP and its significance in the AI landscape.
2️⃣ Azure Language Services Overview: Features and tools for NLP tasks.
3️⃣ Step-by-step guide to:
Text analysis (sentiment analysis, key phrase extraction, etc.).
Named Entity Recognition (NER).
Language detection and translation.
Conversational AI capabilities.
4️⃣ Real-world applications of NLP in industries like customer service, healthcare, and e-commerce.
🛠️ Who Is This For?
Beginners curious about Natural Language Processing.
Professionals preparing for the AI-900 exam.
AI enthusiasts and developers looking to integrate language intelligence into their applications.
📌 Key Features Highlighted:
✅ Build intelligent apps using Azure AI NLP tools.
✅ Real-time text analysis and language understanding.
✅ Simple, no-code/low-code solutions in the Azure portal.
✅ Scalable NLP services for enterprise-grade projects.
💡 Take your understanding of NLP to the next level with Azure AI!
Explore Our Other Azure Courses and Practice material on: https://www.youtube.com/@skilltechclub
Category
🤖
TechTranscript
00:00hi guys good morning good afternoon or good evening from whichever place you
00:13are watching this video my name is Maruti and I'm here with the daily dose
00:18of Azure and AI today we are going to talk about Microsoft Azure AI language
00:24service now this language service is totally connected with the concept of
00:28natural language processing if you do not know what is NLP or what is natural
00:33language processing I strongly recommend you to go through this video first which
00:38is actually going to show you the concept of natural language processing in this
00:42video we are going to talk about step-by-step process where we are going
00:47to create a language service and we are going to understand some important
00:51practical aspects of question answering with natural language processing
00:55basically if you have seen any particular chat bot kind of application or the most
01:01famous one chat GPT kind of application you know what is the chat bot now if you
01:06want to create similar kind of question answering solution where users are going
01:10to ask questions and your chat bot is going to create an answer or is going to
01:15give an answer from the existing knowledge base then this video you should
01:19not miss let's get started I have already logged into my Azure portal and as usual
01:25the first thing we are going to create is we are going to create our AI service
01:29because we are talking about language service today I am going to select
01:33language service which is an independent language service just for your
01:37confirmation language service is available as a multi-service account
01:41inside as your AI service also but we are going to create it separately because
01:45this video is dedicatedly focusing on language service I'll click on create and
01:50while creating the language service it is first going to show you that some of the
01:55basic features are already part of the service if you want to add additional
01:59features you have to select them now you can see sentiment analysis key phrase
02:04extraction name entity recognition or there are a couple of many more which are
02:08already part of the service by default if you want to develop something which is
02:13focusing on custom question answering then you have to select this
02:16particular feature and same way there are some more features which I'm going to
02:21use in my coming videos and that's why I'm going to select this one also which is
02:25custom text classification custom name entity recognition and custom sentiment
02:30analysis now these features we are not going to use in this video but in the
02:35coming videos I'm going to use them so I'm going to select this feature also and
02:39I'm going to click on continue to create your resource this is a language
02:44service with additional features available inside that let me create a new
02:49resource group for this one so I'm specifying AI 900 RG we are learning
02:55fundamentals of Azure AI region is fine creating a name of the service like
03:00Maruti language service and giving a number which is going to make sure that
03:07this is going to be a little unique I'm going to select standard s1 plan as a
03:11pricing tab because I'm going to use some custom additional features also and
03:15while creating this is actually going to create some custom question answering
03:19search solution which is actually going to help you to search the question
03:24from the knowledge base so this is basically going to use as your AI search
03:28service also if you do not know what is your AI search service well you should
03:32follow this particular video which is going to show you the exact detail about
03:35Azure AI search service and very soon we are going to create a lab on
03:40Azure AI search as of now the service is just going to be used with this
03:45particular language service so we are not going very in-depth inside this to
03:49create a custom test classification we need a storage account so we need to
03:53choose whether we want to create a new storage account or we want an existing
03:57one I'm saying this is going to be my new storage account so I'm giving Maruti
04:02storage AI and giving some number at the end of this the storage account type is
04:08going to be standard LRS which is perfectly fine all other things for
04:13responsible AI we are going to say we agree with the responsible AI and we are
04:17not going to misuse AI and that's it I'm going to click on next all the things are
04:24fine live you plus create is going to pass the validation and once this
04:30validation is clear we are going to click on create so that is going to submit my
04:35deployment and it's going to create my language service okay it's showing me
04:40deployment succeeded so I'm going to click on go to resource group it's going
04:45to show me that language service is successfully created and with that the
04:49search service and storage account is also created in this video we are not
04:53going to talk about search service and storage account but yes language
04:56service is what we are going to use so I'm just directly going to click on
05:00language service it's showing me that my endpoint is successfully created left
05:05side if I click on keys and endpoint is going to show me that we have a keys also
05:10which are available and if I click on features this is that section which is
05:14going to show me that what kind of features we have enabled for this because
05:18we want to use question answering we have enabled custom question answering
05:21solution with this and we are all good to go if this is successfully created the
05:27next thing we have to do is we have to go to our language studio now on the home
05:32page itself you can see there's a link here which is showing you get started
05:36with language studio I'm going to click on this and it's going to take me to a
05:40separate tab where it's going to open language.cognitive.azure.com remember
05:45whenever you are using language.cognitive.azure.com the first thing you
05:50have to do is you have to associate this with your resource now I am going to
05:54click on sign in I'm going to choose my account okay once I log in I have to
06:07select my directory my subscription as well as my language service I'm going to
06:13choose my resource type is language service and I'm going to choose my
06:16language service which I have created my service is created with the standard s
06:20this was the name of the service so I'm good I'm going to click on done and that
06:25is going to connect this language service with this language studio remember
06:29one thing whenever you're going to associate your service with the language
06:33studio then only you'll be able to create a language specific projects now
06:37right now it's showing me some of the demos which I can try out but I'm not
06:41going to try a demo I'm going to create a new project from scratch and that
06:45quest project is going to create a custom question answering solution kind of thing
06:50now when I click on this create new kind of a drop down is showing me couple of
06:54options we will try most of these options in our different different demos but as
06:59of now I want to choose this which is a custom question answering solution
07:03remember this is showing you here in this notes that this is going to
07:07customize the list of questions and answers extracted from your content
07:11basically that is what we call knowledge base now this word is very
07:15important guys as I told you that if you have ever used any chat bot chat bot is
07:20actually trained on certain language based machine learning models now this
07:25models are always going to use data in our case the knowledge base word is
07:29actually going to provide that question answer kind of a data whenever you're
07:34going to ask a question to your chat bot from where is going to look for the
07:37answers well the answer is this knowledge base and this custom question
07:41answering solution is actually helping you to create this knowledge base and
07:45customize it now once your knowledge base is ready obviously you have to
07:49associate that with your bot now in that case you have to use a services like
07:54Azure board services we are not going to use as your board right now we will not
07:58develop boards right now but we have this kind of a custom question
08:02answering knowledge base which we will create and then we will try to taste it
08:05whether it's working fine or not that's what our goal let's click on create
08:10custom answering solution it's trying to create a new project it is asking me
08:15certain options like how would you want to select the language for project in
08:19this resource now I can choose a language of a specific language like
08:23English or French or German or maybe I can go with this that I want to select the
08:28language when I create the project in this resource so at that time maybe I can
08:31choose and in this way we are not specifying any specific language as of now so
08:37we are selecting this and we are going to click on next in the basic information is
08:42asking me project name description source language and the default answer when
08:46there is no matching answer found now in all this case we are going to provide
08:51proper information so my project name I am going to take one example from
08:56Microsoft learn only which is on Magis travel Magis travel is a travel agency and they are
09:02actually providing customer support as well as travel related information as well as they
09:07also help in booking hotels and trips now in this kind of scenario Magis travel is
09:13actually providing end-to-end solution for your travel and we have some kind of a
09:17knowledge base kind of a word documents which we are going to use as a knowledge
09:21base and based on that particular document this is going to find answers of the
09:26questions which your customers are going to ask description of the project it can be
09:31anything I'm just saying it's a simple knowledge base the source language I'm
09:35going to choose English because I'm going to ask all the questions in English and
09:38in the default answer when no answers are found we are going to say no answer
09:43found I think this is fine we'll click on next review and finish is just showing me
09:48the same details which I've selected I'm happy with this I'm going to click on
09:52create the project creation is successfully done I am able to see 13 sections here the
10:00first section is showing me manage sources then we have edit knowledge base then we
10:05have deploy knowledge base and then once you deploy knowledge base it will be
10:10available on a public endpoint so you can use it now as of now we will do this
10:15three steps one by one first one is manage resources so let me just click on
10:19the add sources section and it is showing me that I can add sources from
10:23various sources locations I can maybe add sources from a URL if you have a
10:28website with the FAQ kind of a page you can put a page URL also in this and that
10:33URL is going to captured by the service and it's going to take all the questions
10:37and answers from that or maybe you want to do it with the help of file or maybe
10:42you just want to take it from the existing sample chit chat sources where
10:46they are just going to give you a easiness in configuring your chat bot for a
10:51friendly or a professional conversation now in our case we are going to choose URL and
10:57then when we add URL I'm going to specify that URL name is going to be
11:02Margie's knowledge base Margie's KB and in the URL I'm going to provide a URL of a
11:08specific document now you can see right now this is Margie's FAQ dot doc X this
11:14is that particular document which is available on a github repository on
11:18Microsoft learn website so this is one of the document which you can use without
11:22any policy violation or something like that we don't need any licensing all the
11:27dummy content is available on this so this is my one knowledge base URL which I've
11:32added if you want you can add multiple URLs also and it is going to add the
11:36questions from all this knowledge base I'm going to click on add all it's going
11:41to add my first source inside this now my knowledge base is successfully added and as
11:47you can see here this is showing me that source type is URL and this is
11:51successfully added now if I go to edit knowledge base this is a section where
11:56it's going to show me how many questions are added from the document now you can
12:00see there are eight questions answer pair which is added now there are one two
12:05three four five six eight questions which are there if someone is going to ask a
12:09question like how can I book a flight that question is available here and they
12:13have an answer of that also when they are giving an answer with that they also
12:18have an alternate question like someone can also ask how can I book a flight or
12:24some other words like can you help me with booking a flight now you can add
12:29multiple alternative questions also in some of these questions when you check you
12:33will have one or more alternative questions now in my case maybe this is
12:38going to be my knowledge base and I want to add one more question or answer
12:43inside this which I can do surely now I'm just going to add let's say one more
12:50question so I'm going to add a new question answer pair eight questions are
12:53already there I'm saying this is going to be my question source is going to be
12:58this one my question is is it safe to travel this month now maybe when I ask this
13:10question I just want to check the current details and then I want to answer that
13:14thing but I'm not putting that logic right now I'm going to say that yes it's
13:21safe to travel but at your own risk so it's like we are giving a diplomatic
13:30answer it's a stupid answer but yeah whatever this is just I want to prove
13:35that my question is also something which I can add inside that so it's saying yes
13:39it's safe to travel but at your own risk done and the moment this question is
13:47there now you can see that now we have total nine questions which are added here
13:51I'm going to click on save and it's going to save the changes now this way
13:56if I want to add more questions I have to add more URLs basically more knowledge
14:01base and then my my knowledge base is going to be bigger if I go back to my
14:05manage sources I can also add chit chat chit chat is going to add friendly and
14:10some chit chat kind of a questions answer pair inside that now let me click on
14:14chit chat and I'm going to say that I want to make sure that when anyone is going
14:18to ask a question chit chat is going to reply with a friendly manner and then I'm
14:23going to click on add chit chat this is also going to add number of questions
14:27answer which are mostly common conversational kind of a question
14:31answers like let's say some customer is asking hey how are you doing then this is
14:35going to answer yeah I'm doing fine now this kind of questions are added with the
14:39help of this chit chat friendly and now if I go back to my edit knowledge base it's
14:44not going to be a nine question answer pair it's actually 99 question answer
14:48pair it means chit chat has actually added so many questions and answers inside
14:52this with multiple alternate questions like let's say someone is asking this
14:57kind of a stupid question that are you a woman and there are a couple of
15:01alternative questions which are there like are you trans you're queer do you have
15:06a gender are you so these are all questions which are nonsense kind of
15:11question and then we have one answer for that that that doesn't really apply to
15:15me this is the same answer which they're going to give which you can also edit if
15:19you want so this is all it is we have created a knowledge base and if the
15:25knowledge base is having multiple questions obviously more questions more
15:29data is better now let me just check whether this is working or not and in
15:34order to do that I'm going to click on this small tiny I can hear this is I can
15:38which is actually for tasting whether this is working or not I'm going to click
15:42on taste and first I'm simply going to say hello if replying with hi which is
15:50great I'm going to say how are you doing today
15:59I think great thanks for asking and now I'm just going to ask some questions like I
16:06want to book a flight can you help me now I know this is not exactly matching
16:14with that question but it has a phrase which is I want to book a flight so when
16:19I ask this question is saying yes our agents can help you to book a flight
16:22between any major airports or any other major airlines we offer competitive
16:26fares for all the flight classes now this is something which is an answer which is
16:31coming from that particular question and it's also giving me a link of a
16:34margistravel.com which means that this is coming from the margistravel knowledge
16:38base now same way whatever questions whatever answers which we have added we
16:44can add it here now I'm going to ask is it safe to travel this month I'm asking
16:54that nonsense question which I asked and which I have added from my side and he's
16:58saying yes it's safe to travel but at your own risk so the point is my custom
17:03questions are also there my questions from margistravel knowledge base is also
17:08there and my chit chat kind of questions are also there you can build this kind of
17:11knowledge base and based on that you can prepare your questions and answer pairs
17:16now I'm good I'm happy with this particular taste and then we can also do a
17:21lot more customizations with this which we are not doing right now but if this
17:25knowledge base is ready the next thing which you have to do is you have to click on
17:28deploy knowledge base when you go to deploy knowledge base this is giving you
17:33a deploy button it is as simple like this that if you are happy with your
17:37knowledge base and if you want to use it by your developers or if you want that
17:42others can use your knowledge base then you just have to click on deploy they're
17:46asking me do you want to deploy this project I'm saying yes deploy margistravel
17:50once deployed you can integrate this with a bot so this is actually something
17:55which is a next step logically so only after deployment you can integrate that
17:59with the bot I'm going to click on deploy okay it's like showing me
18:05successfully created a deployment so you can see right now deploy knowledge base
18:09and create a plot in few clicks is what you can do right now now we have just
18:14now deployed this and because the bot is deployed is saying your knowledge base is
18:18deployed you can get your prediction URL or create a bot now when I click on this
18:22get prediction URL this is actually going to show me one URL this URL is live
18:28right now and it is hosted on the Azure cloud if you want to check when you make
18:32a request to this it is working or not you can actually just send the post
18:37request using a curl this is a full URL of the curl which they have given here
18:41while you are sending a post request you only have to take care of your question
18:47here so you can see they have given you one parameter that you put your question
18:50there then you put your score threshold here and then you can just specify what
18:56kind of configuration you want to do in this now if you do not want to put a
19:00score threshold it is fine you can add additional property values if you want but
19:04for me the question is something which is important here if you put this thing if
19:08you copy this particular request and if you just do it in the command prompt you
19:13just send the post command it's going to send the request to this API and it's
19:17going to give you the response now this is one of the easiest way to check whether
19:20your service is up and running it's live or not now I'm sure this is actually live
19:25so I'm not going to do that thing right now also if you are happy with this
19:30particular knowledge base if you want to use this API and if you want to create a
19:34chatbot you actually have to click on this button which is create a bot now
19:38remember I'm not going to click on this create a bot right now because when I click on this
19:43it's actually going to use azure bot service using that azure bot service azure
19:48bot framework we are actually able to create a bot in few seconds actually but
19:53I'm just not going to click on this right now because I do not want to create a
19:56bot I have already tasted my knowledge base in the testing section and it is
20:00working fine