Abstract: In order to solve the problem that the important words of short text are ignored when recognizing the named entities of online shopping comments, by referring to the multi-head attention mechanism, contribution of important vocabularies, and 
bidirectional long-term and short-term memory conditional random field model, a named entity recognition method for online shopping comments based on MA-BiLSTM-CRF is proposed. Firstly, the combination of word vectors and part-of-speech vectors is used to represent the semantic information of the comment text. Secondly, the bidirectional long short-term memory (BiLSTM) network is used to extract text features. Then, the multi-head attention mechanism is introduced to improve the model performance from multiple levels and perspectives. Finally, the named entities are identified based on conditional random field (CRF). Experimental results show that this method can improve the recognition effect of online shopping review entities.

Key words: named entity recognition, bidirectional long short-term memory, multi-head attention mechanism, conditional random field, deep learning