Preview

Start a free Courses trial
to watch this video

Sign up for Treehouse

Text Analytics API Part 2

8:42 with Carling Kirk

We'll add serialization logic and make our request to the Text Analytics API.

Teacher's Notes
Video Transcript
Downloads

Parsing the Sentiment Score to a Double

If you're not located in the United States, you might need to call an overload of the double.TryParse method that allows you to pass in the "en-US" culture info in order to successfully parse the string value.

double score;
if (double.TryParse(sentiment.Score, NumberStyles.Any,
    CultureInfo.GetCultureInfo("en-US"), out score))
{
    newsResult.SentimentScore = score;
}

For more information about how to get and set your current culture in .NET, see the MSDN documentation page for the CultureInfo.CurrentCulture property.

https://msdn.microsoft.com/en-us/library/system.globalization.cultureinfo.currentculture(v=vs.110).aspx

Formatting Numbers

Standard Numeric Format Strings

Project Ideas

Build an app that integrates with data from a public API.
- Check out this list of public APIs and find something you're interested in!
- Use other API services from Microsoft Cognitive Services
Download CSV or JSON datasets from Data.gov and load it into your application

We need to serialize our request data into JSON and 0:00

then send it to the text analytics API URL. 0:03

I wanna to show you a quicker way to use JSON.net to serialize objects 0:07

using the JSON convert class. 0:11

You might have seen it earlier in the JSON.net documentation. 0:13

String request Json = 0:17

JsonConvert.SerializeObject and 0:20

will send it the sentimentRequest. 0:26

There's one small bug I found with our sentiment request class. 0:34

Let's go fix it before we build our request. 0:38

This JSON property needs to be documents. 0:41

Let's go check the JSON response in the docs to make sure it matches. 0:44

There it is, okay, let's get back to our method. 0:49

We need to convert our request based on into a byte array so 0:56

we can pass it to the upload data method. 0:59

Remember when we talked about encoding? 1:02

It's a good practice to use UTFA for sending data over the Web. 1:04

UTFA is the most compatible encoding. 1:08

Let's convert our string into a byte array using UTFA. 1:11

To do that, we can use the Encoding 1:16

class by RequestBytes = 1:21

Encoding utf8.GetBytes, 1:26

and we'll pass it. 1:31

Our requestJson. 1:34

Then we can send it to the URL in our upload data method. 1:37

But our URL is still the old one. 1:43

Let's delete that. 1:46

And we'll put a string placeholder and paste in request bytes. 1:48

We need to go grab the URL from the documentation and 1:53

paste it into the upload data method. 1:55

Here it is. 2:01

Copy and paste. 2:02

Now, we can use the json convert class to deserialize the response. 2:08

But we need a string for that. 2:13

Json convert only deals with strings. 2:14

String sentiments. 2:19

equals encoding dot UTF8 dot Get String. 2:21

And we'll pass it our response. 2:29

And now we can use the deserialize object method on Json convert. 2:31

Sentiment response equals Json convert dot deserialize 2:36

And our type parameter is sentiment response, 2:45

and then we need to pass it the string of sentiments. 2:51

All right looking good. 2:56

Now we need to do something with the sentiments from the news headlines. 2:58

We can add a property to the NewsResult class 3:02

to store the sentiment scores from our text analytics response. 3:05

Let's take a look at the response JSON again. 3:09

The score field has decimals, so which type do you think we should use? 3:14

How about a double? 3:18

It doesn't need to be super precise. 3:20

New search class and 3:25

NewsResult public double 3:28

SentimentScore get set. 3:33

Now let's go back up to main In our program class. 3:38

Let's see where we can call this method. 3:44

How about right after we call GetNewsForPlayer? 3:46

SentimentResponse sentimentResponse 3:51

= GetSentimentResponse And 3:58

we'll pass it our list of news results. 4:03

Then we can loop through the sentiment responses. 4:07

So for each var sentiment in 4:11

sentimentResponse.Sentiments Cause it's a list, 4:15

and then let's add another loop to find the correct news result and 4:25

update the score, foreach (var newsResult in newsResults). 4:30

Will add a check to see if the headline matches the ID of the sentiment and 4:41

then we'll add the score to the news result object. 4:46

So if the news result headline. 4:49

because if you remember we put the Sentiment.id equal to the headline. 4:55

Then we'll add the score to the news result object equals sentiment.score. 5:04

But we need to parse the string from sentiment into a double 5:14

We can use good old Double.TryParse. 5:17

We'll need a double to hold the value. 5:20

If Double.TryParse, and 5:23

in we'll pass it sentiment.Score, and out score. 5:29

And so if that's successful, 5:38

we'll assign newsresult.sentiment score equals score. 5:41

We can get rid of this line right here Okay, and since 5:49

we found a match we can break out of this loop to continue to the next news result. 5:56

Now, let's add the sentiment score to the text that we write to the console, 6:02

Sentiment Score and we'll pass it a 0. 6:07

Since it's at the beginning, and then we'll need to change all these that 6:12

follow, One more 6:16

in the list here. 6:22

So we've got 1, 2, 3, 4, and then we'll need to add the sentiment score On 6:27

the result object, result.SentimentScore, 6:32

and I like to double check the numbers, we go one, two, three, four. 6:38

One, two, three, four, great. 6:42

Let's run it and see what we get All right. 6:47

It looks like our sentiments are there, 6:53

but do you think we should format them a little better? 6:55

The scores are supposed to represent a percentage. 6:59

We can use a numeric format or in our string. 7:02

format method. 7:04

The letter P. 7:05

will format the numeric value into a percentage. 7:06

Stop debugging. 7:10

So a lot like our date formatter, right here we just do a :P. 7:13

Now let's see what we get. 7:17

All right, great. 7:23

It's a percentage. 7:25

You can find more ways to format numbers with the link in the notes. 7:27

It's a lot easier to read. 7:31

Let's see, this is two percent, missing the injured Marco Verratti and 7:34

Claudio [LAUGH] Struggled, 7:38

yeah, I guess that's a low sentiment. 7:43

Let's look at some other ones here, Looking for low ones 7:45

These are all pretty high, 9%. 7:54

Looks like somebody had hand surgery. 7:57

Suffered the injury. 8:02

Yeah, that's pretty low. 8:04

All right, great Fantastic job. 8:05

You learned a lot in this course. 8:11

We learned about file IO and how to read and write files to your computer. 8:13

Then we explored how to parse text data into dates and numbers. 8:17

We use Json.net for serialization and then send 8:21

it over the Web Now you can take what you've learned and build some projects. 8:24

Go find some JSON data that interests you and build an app around it or 8:29

better yet find an API and use it to grab data from the web. 8:33

I've put some project ideas for you in the notes. 8:37

Go forth and build things. 8:40

You need to sign up for Treehouse in order to download course files.

Sign up