Google uses artificial intelligence to develop a tool that allows talking to books

Google today launched two new experiments using artificial intelligence to allow web users to use phonology and natural language processing to talk to e-books, using advanced artificial intelligence technologies to enhance their goals and to create programs that can understand and analyze human language components.


Google has combined the two experiments into a single site called Semantic Experiences, which contains interactive tools for natural language processing using artificial intelligence. The site has two experiments: Talk to Books, and Semantris.

The first experience is called Talk to Books. It's a completely new tool for exploring books. You're asking a sentence or a question to search for it in more than 100,000 books, not relying on the traditional way of matching keywords. You can talk to books and get answers that can help you. Whether you're interested in reading it or not, the answer phrase is displayed in bold, along with some of the text that appears next to the sentence in the same context.

This feature is unique and can help you find interesting books that may not appear in keyword searches because they search inside your books for specific sentences instead of words, such as in Gmail's smart response. If Talk to Books does not find On the responses you like, you may get better results with different words or words, and the experience is often better with whole sentences instead of just keywords or short phrases.

The second is called Semantris, a game that tests your proficiency in vocabulary and links to each other and is supported by the same technology used in the Talk to Books tool. Every time you enter a word, artificial intelligence technology is used to look at all the words in the game and choose what you think It is more relevant to this word, and because artificial intelligence technology has been trained in the conversational text, which encompasses a wide variety of subjects, it is able to collect many related words.

During Semantris, when Artificial Intelligence sorts the list, the most related words are moved down. For example, if you were given the word "bed" at the top of a 10-word set, you might think you should type "sleep" as the answer. Semantris will then classify the 10 words and give you points based on their assessment of the semantic relationship between bed and sleep compared to the word bed with the words in the list.

The game can be run with the first two versions of the list, which is called Google Arcade, and it needs to think and write quickly so that you can collect a lot of points, and the second is the version of Blocks Blocks, which allows you to enter the appropriate words to clear the blocks of the screen based on your assumptions associated with words written on the blocks Colorful.

It is worth mentioning that Google has developed the understanding of natural language dramatically in the last few years partly due to the development of the word vectors model, which enables algorithms to recognize the relationship between words based on examples of actual language use. These vector models draw language-like expressions for nearby points based on the equivalence, similarity, or connection between ideas and language.

With such experiments, Google has been able to demystify artificial intelligence technology and terms such as computer learning and neural networks in a simplified way for everyone, and have made their application in fact easy and user friendly.