Mongolian multi-modal fine-grained sentiment analysis method fusing prior knowledge model

Abstract

The invention discloses a Mongolian multi-modal fine-grained sentiment analysis method fused with a prior knowledge model. The method comprises the following steps: preprocessing a Chinese and Mongolian sentiment corpus containing emoticons; converting the text words and the emoticons obtained through preprocessing into dynamic word vectors; introducing an attention mechanism to carry out fine-grained sentiment analysis on a word vector spliced by the text word quantity and the emoticon word vector; respectively creating a Mongolian emotion dictionary and an emoticon emotion dictionary, and taking features extracted by the Mongolian emotion dictionary and the emoticon emotion dictionary as emotion features finally extracted by the model; pre-training the model fused with the prior knowledge on large-scale Chinese and Mongolian linguistic data to obtain a Mongolian multi-modal fine-grained sentiment analysis model fused with the prior knowledge model; and comparing and evaluating the analysis result of the model and the analysis result of a single network analysis method according to the accuracy, the precision rate, the recall rate and the F1 value of each emotion category, so that the purpose of improving the Mongolian text emotion analysis performance is achieved.

Description

technical field [0001] The invention belongs to the technical field of artificial intelligence, in particular to a Mongolian language multimodal fine-grained emotion analysis method fused with prior knowledge models. Background technique [0002] With the rapid development of Internet technology, more and more people begin to express various opinions on social platforms such as Weibo, forums, film and television websites, shopping websites and other platforms to share their moods, views and opinions. In particular, with the rapid development of network technology, emoticons have gradually developed into a new data form different from text, images, and videos, and play an important role in the field of sentiment analysis. The texts, emoticons and other information published by users may contain different emotional colors: happy or favorite; sad or angry. The core of sentiment analysis is to accurately divide the emotion expressed by a piece of text into seven categories: hap...

AI Technical Summary

Problems solved by technology

First, due to the rich and varied forms of Mongolian vocabulary, there are serious unregistered words in the process of Mongolian text sentiment analysis, and the existence of a large number of unregistered words seriously affects the accuracy of sentiment analysis.

Second, current sentiment analysis mainly analyzes text data. For data containing emoticons, important information such as emoticons is generally deleted during the data cleaning stage, which cannot fully utilize emoticons, an important feature of sentiment analysis.

Third, the current single neural network model has limited classification efficiency when solving sentiment analysis, and is limited to certain feature data information

Images