Research On Semantic Expanded Search Engines

With the rapid development of computer network technology and the influx of electronic texts, search engines have popped up and boomed in order to satisfy users’demand for information. However, the results returned by traditional search engines are not quite satisfactory, so users still need to spend a lot of time screening the search results to find what they are really interested in. The booming Semantic Web and accumulation of semantic information enable the search engine which is based on semantics to make full use of semantic knowledge. Thus it can process semantic extension reasoning on users’query and automatically summarize query results returned by the search engine based on semantic text. Semantic search engine can provide more accurate and useful information for the users. Meanwhile, it reduces users’workloads and turns out to be more time saving. As a result, the study of semantic search engine will promote the development of search engine and have an important significance on Internet technology.On the basis of existing semantic search engine, this thesis proposes a complete and expanded framework for semantic search. The framework contains eight parts, they are query pre-processing, interface engine, ontology servers, ontology query expansion, information collection and processing, indexing and indexing database, automatic summary, the results of sorting and display.This thesis proposes an algorithm on solving the problem of huge scale of query words expanding and noise data. On the basis of ontology, system constructs a semantic tree, and then calculates the semantic distance, last provider a reasonable extension query words to users.Moreover, the mechanical extraction algorithm of keywords is improved for automatically summarizing the semantic text in this paper. We add relevant methods based on the semantic summary extraction and put forward the methods of calculating the importance of sentence, determining the summary sentence and eliminating. The method has higher accuracy recall summary statistics than the traditional mechanical method.Finally, the sorting of search results is also explored. We introduce the traditional ranked positional method into abstract sorting, which turns out to be effective in the actual application. This method is based on semantic comprehension, integrating position sorting method, which is more accurate, reasonable and effective page sort method.