Is the data analysis borderless? (new Google APIs).

 Blogs
Is the data analysis borderless? (new Google APIs)

Do you miss the proper analysis of the data from pictures, videos or long texts? No problem. A few weeks ago Google introduced its new APIs, which solve a lots of problems and gives the answer to a lots of questions.

Google Vision API

DEMO: https://cloud.google.com/vision/

This API gives us the power to analyse pictures and the essential information which they contain.

To know Google Vision API better I tested it a bit. I uploaded a photo made on the top of the Eiffel tower. Google Vision API, diagnosed the photo. It correctly determined the place where it has been taken (Paris, Eiffel tower) and GPS coordinates.  

Furthermore, the photo was given labels, with the probability the label is correct. These labels – categories, can help us by the analysis of big amount of pictures, to be able to find out which pictures are most shared. We can also analyse if there is some pattern, and which type of images engage our audience the most.

In my case the photo was categorized as follows:

API can also evaluate, if there are some similar pictures on the internet. It scans also the picture itself and says what is on it.

In my test:

Google Visual API can also evaluate, if the content of the picture is humiliating, inappropriate for children or shows any violence.

In the case, the picture contains face on it, the API can evaluate the emotions of the person on the picture. If the picture contains any text, the API can read it and gives us this information as well.

Practical usage

Let’s assume, we have a blog with thousands of readers, where all the registered users can add an article with her/his own pictures.

API can help us to analyse the pictures from the copyright point of view – if the picture is an original, or it is downloaded from Google, without any permission to post it.

We can also analyse an adult-only content and so it can potentially save our reputation.

Deeper research can help us to determine all the categories, which have the highest influence on website visitors, and how these visitors are converting.

Google Natural Language API

DEMO: https://cloud.google.com/natural-language/

Google Natural Language API helps us with texts. It analyses the relationships between words, and each words is categorized by manners of an entity and an importance.

Example – I copied the text into the API. And this was the result:

Natural Language API analysed the text and I indeed saw the words, its entities, and categories, where the words belong to.

By colors it determines which words are most important ones and which give an real value. It creates a bullshit map”. More grey – bigger bullshit. The more colors – more valuable text.  

Very interesting is the document & sentence level sentiment analysis.

This shows us how sentimental is the text, and we can analyse the readiness of the sentimental and non sentimental articles also in this way.

Practical usage

Over this API we can analyse the whole blog articles, the description of an ecommerce product and their influence on the readers, and conversions. This can also help us by identifying the hoaxes. It depends on us if we focus more on sentiment analysis or on bullshit analysis.

Google Video Intelligence API

DEMO: https://cloud.google.com/video-intelligence/

Very similar to Google Vision API is also Video Intelligence API

It can recognise the objects in the video, times in which the object appears, and the probability the API is correct.

It can also estimate, where the video was taken, or it can analyse the text/ speech in the video.

Practical usage

On our blog, where all the users can contribute with their articles, might be also a lots of videos. We can analyse these, subtract text and estimate, which type of video is receiving the highest amount of traffic.  

We can also analyse all the object in the video, and the times where someone is talking. The analysis of the colors can help us to answer the question: Are the darker videos more interesting as the light ones?

Imagine, on our web site there are thousands of videos. As a admin of the web page, you do not have time to watch all these videos and to evaluate them.

But with the help of an analysis and the statistics we can find out exactly what is happening and when it is happening. And if you know, what videos are most engaging, you know, which topics are interesting for your audience as well.

And if you know, your audience, you can offer better services.

Technical output

Technical output from the new Google APIs is in JSON format, so it is easily accessible for the next data analytics in Python, R or REST API. So it is up to you which tool you prefer the most.

There are definitely much more possibilities, where to use these APIs. And it will become more interesting, when we tie more APIs together.

New Google APIs:

 

Leave a Reply