Cross-Lingual Tense Tagging Based On Tree-Structured LSTM

In today ’s increasingly globalized world,people from different countries and cultural backgrounds communicate more and more frequently.Chinese English machine translation has always been a hot research direction.With the rapid development of deep learning in recent years,neural network machine translation has made a great breakthrough compared with traditional machine translation methods.However,it is still far from the requirement of high standard "faithfulness,expressiveness and elegance".The tense problem of machine translation is one of the problems that has not been completely solved.The first work of this paper aims to solve the problem of temporal processing in the process of Chinese English machine translation.Combined with previous work,this paper proposes a temporal annotation algorithm based on Tree-LSTM.Tree-LSTM is a kind of graph neural network,which improves the standard LSTM and makes tree LSTM accept the data of tree structure as input.In this paper,we use tree LSTM as temporal annotation,combined with word vector and temporal information features,and manually annotate 200 Pennsylvania treebank files as experimental data.The experimental results show that the temporal annotation algorithm based on tree LSTM has a certain improvement compared with the previous research results,which shows that this method has a certain feasibility.The second work of this paper constructs the semantic mapping from WordNet to HowNet.HowNet is one of the most representative lexical knowledge bases,which covers a wealth of semantic knowledge.In recent years,more and more scholars in the field of natural language processing focus on how to integrate the semantic knowledge of HowNet into the model.This paper proposes a semantic mapping algorithm from WordNet to HowNet,so that English words can also be represented by sememes and def defined in HowNet.