Recommendation method based on explicit information coupling analysis of bidirectional long-short-term memory network

Abstract

The invention discloses a recommendation method based on explicit information coupling analysis of a bidirectional long-short-term memory network. The recommendation method comprises the steps of dataacquisition and processing, data set division, coupling model construction, model training and project recommendation. According to the method, a BiLSTM bidirectional long-short term memory network and an Attention attention mechanism are combined by analyzing a microscopic coupling relationship between explicit short text information related to a user and a project to obtain explicit text feature representation fused with important context information, and the explicit text feature representation is combined with implicit features to obtain an explicit text feature representation model. Thecoupling relation between the comment text information and the subjective emotion of the user is analyzed, the subjective preference of the user is better expressed, the explicit and implicit couplingrelation of the user / project characteristics is fused, and more accurate personalized recommendation is provided for the user. Meanwhile, the convolutional neural network is adopted, so that interaction between learning features at a deeper level is facilitated.

Description

technical field [0001] The invention belongs to the technical field of natural language processing and computer artificial intelligence, and in particular relates to a recommendation method based on explicit information coupling analysis of a two-way long-short-term memory network. Background technique [0002] Natural language processing usually uses pre-trained word vectors to complete subsequent tasks. The initial word vectors are obtained through unsupervised training through a shallow network. Although they show good characteristics at the word level, they lack the intrinsic properties of continuous text. Expressiveness of relational and linguistic features. Nowadays, comment text information has gradually become an important research basis for recommender systems. However, many existing recommender systems have poor interpretability, and consider that users and items are independent and identically distributed, ignoring the coupling between users and items. At the same...

AI Technical Summary

Problems solved by technology

Nowadays, comment text information has gradually become an important research basis for recommender systems. However, many existing recommender systems have poor interpretability, and consider that users and items are independent and identically distributed, ignoring the coupling between users and items. At the same time, there is no Achieving the combination of sentiment analysis of review texts and non-independent and identically distributed recommendation methods

Images