BPT AI Deep Learning Structure, Framework

AI deep learning of patent document

Since the claims are the main content of the patent document, the rights are guaranteed by the content written in the claims. However, the analysis of these documents is very difficult to read and contains a lot of content. BPT AI carries out the following preprocessing steps to analyze these claims. They are converted into RDD (Resilient Distribution Dataset) to be analyzed in Apache-Spark, divided into words, and stop words are deleted. RDD stands for Spark's basic data structure.

Word2Vec Analysis Logic

The search equation is composed of the input information and operators, and it is difficult to obtain the exact information needed. Word2VEC is a technology that extracts accurate data by vectoring words and measuring their degree of similarity. The contents of the patent document are schematized more accurately, and irrelevant information is removed by the NOT operation.