Overview of Cognitive Services

There are different Cognitive Services that you can use by receiving a web service URL and a key. To check out the available services, you must navigate to the Cognitive Services web site [1]. As you can see in Figure 18-1, there are five main categories for solving business problems: Vision, Knowledge, Language, Speech, and Search.

As you can see in Figure 18-2, there are five different APIs for business language problems. Text Analytics is one of the most popular features for language detection, identifying main keywords in text, topic extraction, and checking how much text is positive or negative in context (sentiment analysis).

The other services in the Language category are about translating a text, identifying the primary intent, and objectives in conversations and texts.

These features can be combined. For example, it is possible to integrate the Language Understanding service with the Speech service, for instant speech-to-intent processing, and, with a Bot application (Chapter 19), to develop a more sophisticated application.

In this chapter, you will see how you can perform text analytics for the purpose of detecting main keywords in customer feedback, how much of the feedback is positive, detecting the language in Power BI, by using Text Analytics services.

First, we must set up the environment in which to use Cognitive Services. To use Cognitive Services, you can sign up for a free trial. The trial services are free for seven days and do not require a credit card. The other way to access Cognitive Services is to use a free Azure account for a month, which provides a $280 credit on Azure, and all data and customization will be saved. The final approach is to use an existing Azure account (Figure 18-3).

To set up Cognitive Services in Azure, you must log in to your Azure account and search for the services you want. After logging in to the Azure portal, you must create a Text Analytics service. To create a Text Analytics service, you must create a new service in Azure, by clicking the top left of the page. In the New page, the main service categories will be shown (Figure 18-4). AI + Machine Learning is one of the main services in Azure. Click it to see the available options. As you can see in Figure 18-4, Computer Vision, Face (recognition), Text Analytics, Language Understanding, Translator Speech, Bing Search, and Azure Search are among the main topics under AI + Machine Learning.

Text Analytics Services

Not all collected data is about numbers and structured data. To gain a more holistic perspective about a customer, products, and so forth, collecting and analyzing the text can be used, to enhance company performance. Text Analytics is the process of converting unstructured text data into meaningful data, to understand customer needs and feedback [2].

In this section, a case study will be presented that explains the process of using the Text Analytics API in Power BI and is applied to customer feedback.

The first step is to set up a Text Analytics service in the Azure portal (Figure 18-5).

There are different pricing tiers for this service. The first is a free tier that allows users to apply the Text Analytics service on 5K rows of data a month (Figure 18-6).

After creating the Text Analytics service, we get the API URL and key access from the created service. Figure 18-7 shows how to access the URL and the access key in the Text Analytics service.

After creating the service from the Azure portal, we must call it inside the Power BI Desktop and apply it on the available data there.

Data Set

Fabrikam is a mock production company. It has received e-mails from customers regarding shipping, tech support, and other concerns. The business intelligence (BI) manager wants to perform some analytics on the e-mail the company is receiving from customers, to better understand the main points of customers’ e-mails and to determine whether they are satisfied with the company’s services. You can download the data set from https://github.com/Kaiqb/KaiqbRepo0731190208/blob/master/CognitiveServices/TextAnalytics/FabrikamComments.csv [3].

The first step is to open Power BI Desktop and load the data set, as a Text/CSV file, into Power BI (Figure 18-8).

Then, instead of loading the data, click the Edit option to navigate to the Power Query environment for data transformation (Figure 18-9). As you can see in Figure 18-9, an overview of the data set is shown. There are about 20 rows of data about customer e-mails, names and IDs, e-mail subjects, and comments.

First, we are going to combine the Subject and Comment columns in Power Query, using Merge Columns under the Add Column tab. The general procedure is shown in Figure 18-10.

After creating a new column, we must rename it to Customer Feedback.

Create a Text Analytics Function

It is possible to create a function in Power Query that applies Text Analytics to a specific column in Power Query. To create a function, we must click the whitespace in a query, at the left side of the main page, and choose the Blank Query option (Figure 18-11).

Rename the created query Sentiment Analytics. The next step is to convert the blank query to a function. To do that, we must access the M Query behind the created query, by clicking on the Home tab, then Advanced Editor (Figure 18-12). Here there is an editor that we are going to use to write some M scripts and change the query to a function.

The next step is to add the following code:

(text) => let

    apikey      = "<API Key>",

    endpoint    = "https://<Location of Your Azure>.api.cognitive.microsoft.com/text/analytics/v2.0/sentiment",

    jsontext    = Text.FromBinary(Json.FromValue(Text.Start(Text.Trim(text), 5000))),

    jsonbody    = "{ documents: [ { language: ""en"", id: ""0"", text: " & jsontext & " } ] }",

    bytesbody   = Text.ToBinary(jsonbody),

    headers     = [#"Ocp-Apim-Subscription-Key" = apikey],

    bytesresp   = Web.Contents(endpoint, [Headers=headers, Content=bytesbody]),

    jsonresp    = Json.Document(bytesresp),

    sentiment   = jsonresp[documents]{0}[score]

in  sentiment

As you can see in the preceding code, the first line is about the function input which is text. In line 2, the API key collected from Azure Service must be pasted here. In line 3, the required URL for connecting to the API must be provided. As you can see at the end of the URL, the service that we are going to use from Cognitive Services is shown. In this example, we are using sentiment analysis. Lines 4 to 11 indicate the required code for connecting to Cognitive Services and getting the result in JSON format. The last line is the result column that shows the sentiment score for each comment.

Now, replace the preceding code with the previous one and click OK. The blank query will change to a function type, and a page with a text box will appear. You can test it by writing a sentence, such as “the weather is so nice today,” and click the Invoke button, to see the result.

As you can see in Figure 18-13, the result of sentiment analysis is 0.92.

The output of the sentiment analysis is a number from 0 to 1. A number closer to 1 means that the comment is positive, one closer to 0 means the feedback is negative.

For a final step, we can apply the function to the customer feedback (Comments) column. Click the Fabrikam data set, the Add Column tab, and then Invoke Custom Function. On a new page, choose an appropriate name for the new column, then choose the function in the drop-down. From the last drop-down item, choose the column you want to apply the function to (Figure 18-14).

A new column will be shown at the end of the data set overview that shows the numeric result for sentiment analysis. As you can see in Figure 18-15, the number varies from 0 to 1. You can click the Home tab and Close and Apply to see the result in Power BI Desktop.

The process of using the other features, such as language detection and key phrase extraction, is the same. Only the URL end point will change slightly. For language detection, the end point URL will change as follows:

    endpoint    = "https://<Location of Your Azure>.api.cognitive.microsoft.com/text/analytics/v2.0/languages",

For key phrase extraction, the end point URL would be

endpoint    = "https://<Location of You Azure>.api.cognitive.microsoft.com/text/analytics/v2.0/keyPhrases"

In addition, the last line of code will change to show a different variable. For language detection, this would be as follows:

    language    = jsonresp[documents]{0}[detectedLanguages]{0}[name]

in  language

For the key phrase, it would be as follows:

    keyphrases  = Text.Lower(Text.Combine(jsonresp[documents]{0}[keyPhrases], ", "))

in  keyphrases

The preceding example was about using Cognitive Services in Power BI. In the next section, I will show you how to use the Face (that is, face recognition) service in a Windows application.

Intelligence Application, Face Recognition Services

Another feature in Cognitive Services is the Face (facial recognition) API. This API performs face detection from an image, in addition to emotion detection and finding similar facial features of faces from two images.

It is possible to try the Face API before using it in an application. To see a demo, navigate to the Cognitive Services web site, then to the Vision category (Figure 18-16). Under the Vision category, click the Emotion recognition in images demo. To test the API, browse to a picture and then submit it. After uploading the image, click the Submit button.

As you can see in Figure 18-17, after uploading the image, the API shows the emotion, such as anger, contempt, disgust, fear, happiness, neutral, sadness, and surprise, indicated by a number between 0 to 1. As you can see in Figure 18-17, the only measure that has a higher number is for happiness, and the person in the picture is smiling.

We are going to create a Windows application that is able to detect the image using the Face API in Cognitive Services.

The process of creating the Face API is like that for Text Analytics. First, we must log in to the Azure portal, then search for the Face API under AI + Machine Learning (Figure 18-18) .

After creating the service, click the Key tab at the left side of the page and grab the key (Figure 18-19).

We are going to create a Windows application that employs face recognition [4]. To write the code, you must first download one of the free versions of Visual Studio 2015 or 2017 [5]. Then you must create a Windows application (Figure 18-20). For this case study, I am using Visual Studio 2015.

The next step is to install the Microsoft Azure Cognitive Service Face API in a Windows application. To install it, we must navigate to the Tools tab, then navigate to NuGet Package Manager, and choose the Package Manager Console. Next, we must type in the following code in the console editor (Figure 18-21).

Install-Package Microsoft.Azure.CognitiveServices.Vision.Face -Version 2.0.0-preview

Then we must create the interface for the application. We have to add a button to the screen. To do that, we change the code. Right-click the MainWindow.xaml page, choose View Designer, and change the code to the following (Figure 18-22):

To change the code, click MainWindows.xaml.cs, then add the preceding code to the reference part (Figure 18-23).

If you ensure that all libraries and reference work are in order, you will gain proper access to the Azure Cognitive Service Library inside the .NET application.

Now you must run the code, by clicking the Start button, then click the Browse button and import a picture, to see the description of the image at the bottom of the page (Figure 18-24). As you can see, the software provides a description, such as my age, emotion, hair color, and so forth.

Summary

This chapter presented a brief introduction to the easy-to-use AI tools available in Microsoft Cognitive Services. A brief description of Cognitive Services and how they can be accessed was provided. Then how to use Cognitive Services Text Analytics API in some other Microsoft tools, such as Power BI, was explained. Next, the process of using Cognitive APIs in a Windows application was discussed, and all related codes were shown. In the next chapter, Bot Framework, another tool for creating smart applications, will be explored.

References