A Chinese question-answer retrieval method for knowledge graph based on dynamic programming algorithm

Abstract

The invention provides a knowledge graph Chinese question answer retrieval method based on a dynamic programming algorithm. The method aims at modeling the original question answering process comprising three substeps into the same process in a unified mode, the entity recognition process and the question answer retrieval process are modeled with a same model by integrating thoughts of a state machine in form language, the entity recognition process is modeled into the process of transferring the state of the system from the original state 0 to the state 1, and the question answer retrieval process is modeled into the process of transferring the state from the state 1 to the state 2. The precise and the scope of an answer are guaranteed, and the overall question answering process can be subjected to unified modeling macroscopically. In addition, the target entity transferring process in the question answering process is modeled into the state transferring process in the state machine,a transfer entity is modeled to be in an intermediate state, and a final answer is modeled to be in a termination state. On the basis, the transfer process of all the states achieves decision supporting as a whole through the dynamic programming algorithm, and meanwhile the operation efficiency of the system is guaranteed.

Description

Technical field [0001] The invention relates to a retrieval method, in particular to a knowledge graph Chinese question answering retrieval method based on a dynamic programming algorithm. Background technique [0002] With the development of the Internet and the continuous growth of network information, more and more information can be retrieved from the Internet through search engines. The search results show the characteristics of quantitative data, diversified forms, and comprehensive coverage. On the one hand, it has improved The possibility for users to search for results, on the other hand, makes it difficult for users to quickly and accurately locate the required information. Obtaining useful information quickly and accurately from massive amounts of information is an urgent need of people in the information age. This demand has also promoted question answering systems to become a research hotspot in the current natural language processing field. [0003] There are two mai...

AI Technical Summary

Problems solved by technology

Under actual circumstances, carrying out engineering practice according to this three-step method will expose significant shortcomings: on the one hand, in order to ensure the accuracy of the final output of the system, the accuracy of the methods used in these three sub-steps is very high. requirements, for example, if the accuracy of each sub-step reaches 0.9, but the cumulative overall accuracy of the system can only reach 0.729; The process of adjusting the system to improve the quality of the system is hindered

Therefore, in the existing technology, the accuracy problem in the question answering system and the modeling problem of the question answering process have yet to be solved.

[0007] In addition, in the question answering system of the prior art, multi-hop questions requiring multiple transfers and retrieval

At present, neither the matching method based on artificial rules nor the training method based on deep learning has a practical solution to this kind of problem.

Images