Opinion mining device and intelligent terminal

Abstract

The invention discloses an opinion mining device and an intelligent terminal. The device comprises a word vector model training module used for training a word vector model by using a preset originalcorpus; a to-be-analyzed text acquisition module which is used for acquiring a to-be-analyzed text, and the to-be-analyzed text comprises a plurality of first words; a word vector conversion module which is used for converting each first word in the to-be-analyzed text into a word vector according to the word vector model; a sentence vector calculation module which is used for calculating sentencevectors of the to-be-analyzed text by utilizing each word vector and the corresponding emotion weight thereof, wherein the emotion weight of each word vector is predetermined; and an emotion categorydetermining module which is used for determining the emotion category of the to-be-analyzed text according to the sentence vector of the to-be-analyzed text. According to the technical scheme, the efficiency and accuracy of text sentiment analysis can be improved.

Description

technical field [0001] The invention relates to the technical field of natural language processing, in particular to an opinion mining device and an intelligent terminal. Background technique [0002] Text sentiment analysis, also known as opinion mining (Opinion Mining), refers to the process of analyzing, processing, summarizing and inferring emotionally subjective texts. Text sentiment analysis technology plays an increasingly important role in network marketing, corporate public opinion monitoring, and government public opinion monitoring. Sentiment classification is a sub-task of text sentiment analysis technology. It uses the results of the underlying emotional information extraction to divide emotional texts into several emotional categories, for example, into two types of emotional categories, where the text corresponding to the positive emotional category is positive. Text, the text corresponding to the derogatory sentiment category is negative text. [0003] In t...

AI Technical Summary

Problems solved by technology

Among them, in the process of training the classifier, it is necessary to manually construct a large number of feature engineering, so manual labor is required for a long time, which greatly consumes manpower; on the other hand, the text vector representation constructed in the traditional technology is sparse and easy to Lead to the curse of dimensionality and make the training time of the classifier longer, thus reducing the efficiency of classifying the text to be classified

Due to the lack of semantic information representation of the text in the trained classifier, the accuracy of the result of using the classifier to classify the text to be classified is low

Images