| With the development of the times, users are no longer satisfied with traditional information retrieval to provide simple link. Automatic question answering in recent years has been paid a lot of attention in many scientific research institutions in China and abroad, but a mature Chinese question answering system has not yet appeared relatively, especially the financial field. So it's significant to explore a more efficient and humane way to provide information retrieval services.My research is based on financial field, explored research on Stock Picking for natural language question, and ultimately has completed the system development. The system contains 3 parts: ontology construction, question analysis and Search. Ontology construction uses the Resource Description Language RDF storage framework as a financial knowledge to build ontologies on top of the financial field. It also converts the basic stock information to ontology knowledge base. The current system has completed the construction of top-level financial ontology and basic stock data conversion. Question analysis part convert the natural language into Conditions of semi-structured, it uses the word segmentation and part of speech tagging, word recognition, preprocessing logic of questions and the concept of knowledge with ontology in the semantic analysis to get the true meaning of questions. Questions through the analysis of stock picking will turn words into different models, each mode represents a semantic. Then the search part will turn the semantic analysis into a structured query language SPARQL and query in the ontology library to get the results back to the user.System uses 2000 stock picking questions prepared for the test sample, the knowledge ontology library can cover 97% of the questions, 95% of the questions to get the right results, 94.15% of the questions can get the right answer .All stock selection can response in less than 1 second and the system has released in our company. |
PDF Full Text Request