Research On Semantic Tagging Extracting Method For Text

With the rapid development of Internet and the advent of the era of big data,various of natural language text information acquisition and storage is more and more rich,and in such a large number of text data,quick search and locate the information you need is crucial,so the text semantic tags to accelerate the search process and speed up the text classification application effect is obvious.Based on the above background,this paper proposes two tags semantic text extraction method,respectively,under supervised and unsupervised type environment effect is obvious,the experiment proved that our method on the extraction of semantic tagging effect is very good.Tag information according to whether the use of text data contains tags semantic text extraction method to divided into two categories: the first is the supervision environment semantic tagging extraction technology,this kind of method by using noted in the original text semantic label information,correction method to extract semantic tagging model structure and parameters,and continuously improve the performance of the model by means of iteration,the final will calculate good semantic tagging extraction model into text environment using without annotation,extract the corresponding text semantic tagging information;The second category is the semantic tagging extraction technology in unsupervised environment,this kind of method does not need to use text originally the semantics of the label information,but only rely on a large number of text itself inherent in the semantic structure distribution information and order information,from the core of the text can be derived from the original text semantic word,used as semantic labels the elements in the collection.In solving the problem of semantic tagging extraction,there are two key points: the first is the text of the calculable representation,information in the text data is mainly composed of natural language form,as the representation of information unit each term.But the words and expression but not directly into mathematical model of semantic computation,so find a corresponding text computable representation is a very basic question in semantic tagging extraction task;The second point is the core of the text semantic calculation for existing computable representation of text data available data model is set up,in this model fitting on text information such as the semantic distribution and sequence structure,and then according to the fitting model computed text semantic label information,which is the core of the semantic tagging extraction process.Based on the above two key points,adopting the distributed says Hinton hypothesis,with vector as the term is the basic semantic representation,under different computing environment respectively corresponding semantic tagging extraction algorithm model was established.Based on the natural language model structure,this paper proposes the method of extracting semantic tags based on the long term memory network(LSTM)based on the natural language model structure.In this paper,based on the idea of prototype clustering,this paper puts forward a text semantic label extraction method based on the word item clustering based on the idea of prototype clustering.Experimental results show that these two methods are feasible and effective in their respective computing environments.