Knowledge Searching And Its Core Techniques For Semantic Web

Internet contains a mass of knowledge with a unique openness, and allows people share the knowledge anywhere in the world. Internet has also hidden some information for the user, such as: communication protocols, machine address and the details of the operating system, allowing users to access to web site easily. Anyway, Internet has provided people a platform, by which we can use and release resources anywhere and anytime. The openness of Internet also caused the rapid growth of information resources. Faced to huge information resources, it is urgent to develop a technology to thoroughly interpret the information that the machine can not understand, identify their semantic information, and implement semantic maintaining translation to the logical structure, so we can deal with a large amount of text information resources and use their implicit knowledge. Whether machines can replace people to automatically accomplish the knowledge discovery, knowledge acquisition, knowledge storage and knowledge reasoning on the Web would be a long-term job. The old standards of Web, such as HTML, cannot be understood by the machine; traditional methods of Web searching schema constrained search precision and recall rate; the Web resources expressed in the form of string are not appropriate to knowledge used by users. W3C has recommended Semantic Web as a new specification to express information on the Internet. We build intelligent Semantic Web searching system on the Semantic Web platform, using semantic information to guide searching. At the same time, we introduce ontology to provide the knowledge representation and deduction of Web resources. In the Semantic Web search, "Web Resources" become "Web knowledge", and "Web information search" upgrades to "Web knowledge search".This paper has studied the core technology of Semantic Web search and the system implementation. We focused on five aspects to discuss the core technology, and then implemented a Semantic-based intelligent search system, KS3W, which has comprised the five kinds of technology.(1) Based on the introduction of ontology, consistency of ontology and basic reasoning method - Tableau algorithm, we have given the definition of the incremental consistency checking problem in the ontology. Whether the original ontology and incremental ontology are merged directly is a point of view. A pre-treatment optimized approach is proposed to check the incremental ontology in advanced instead of emerging original ontology and incremental ontology directly to optimize the consistency checking. By checking whether the minor checking sets exists, we have proposed the compact space optimization to check the consistency by defining the terminology minimum checking space and assertion minimum checking space, Further, we have combined the two optimizations, and proposed the incremental consistency checking algorithm based on the preprocessing and space compact optimization. The soundness and termination of our algorithm are proved at the same time. Its evolution is given to compare with the traditional approaches. The result shows that our approach is more effective than the traditional approach under specific conditions. The average efficiency is improved about 15%.(2) At the basis of the analysis and comparison of the current topic-based crawling, we have studied the web crawling which uses ontology to depict domain topic in depth, and proposed the semantic-based focused crawling under the framework of ontology and reasoning of the ontology. Through the establishment of ontology mapping mechanism, we have given the semantic to keywords, and used crawling reasoning task (consistency extend and range relational reasoning under the user-defined focus crawling to construct the semantic overlay effect model. For the concepts in a group of web pages obtained by the effect model, we apply the concepts inclusion relation to determine the topic relevance. By this way, we get the page link with sort out and implement the web crawling. The semantic overlay effect models consist of instance semantic effect overlay, property semantic effect overlay, and peak effect overlay. As the experiment verifies, the crawling strategy in this paper sufficiently applies the ontology semantic information to reason and calculate. The method is superior to other related algorithms.(3) Automatic semantic annotation is an important research field of artificial intelligence problems. This chapter has focused on the study of the semantic annotation on Web resources. At the basis of the analysis of the existing semantic annotation, we have summarized the mainstream methods of automatic semantic annotations. The traditional methods use the machine learning and model-based method to realize the automatic semantic annotation. Though these methods are constantly being improved and upgraded, they are either based on statistical probability or expert experience, which results in the limited accuracy. In this paper, we have proposed that using ontology reasoning to implement the ontology hierarchy stratification, and set the obtained relational concept located in the top level of ontology as the criteria of the machine learning algorithm to identify the entities to be checked and then construct the logical expressions of the instances to be checked to fine the instance concepts. We consider the entity identify from the aspect of the logic level, and supplement the semantic annotation which only considers the statistical probability. At last, we use the named entity recognition, anaphora resolution and other natural language processing technology to generate semantics.(4) Since the traditional storage method did not consider the data integrity, the data mapped to the database may not satisfy the integrity in the relational database. In this chapter, we have defined the IC-Mapping axioms to simulate the integrity constraints in the relational database. In further, with these axioms, the integrity of the original ontology has been checked and the data that didn't satisfy the constraints has been modified. The IC-Mapping axioms were applied to the IC-based storage to optimize the existing storage approaches. Finally, our storage approach was compared with the other approaches- vertical storage and class-based storage approaches. As the result shows, the query efficiency of the IC-based approach is superior to the other two. In addition, with the growth of the dataset scalability, the gap is more obvious.(5) By applying the checking of the concept inclusion relation, we have proposed the semantic reduction optimization of SPARQL query and combined it with the selectivity-estimation optimization. Further we have proposed RS_Opti optimization and given its implementation. The comparison of the algorithm by LUBM Benchmark shows that RS_Opti optimization is superior to the optimization which uses the semantic reduction and selectivity-estimation optimization separately. Additionally, we have compared the RS_Opti optimization with current SPARQL query engines. The result shows that the advantages of our RS_Opti optimization are more obvious, when there are more tuple patterns and complicated semantic relations in the query instance.(6) Since the traditional search engines are limited to understand user queries and the content of the current Web, which limited the improvement of the web search technology, we have developed KS3W, a novel Semantic Web intelligent search system which realizes the Web knowledge discovery, knowledge integration, knowledge reasoning and integration. The design of the prototype system consists of four subsystems, in which the subsystems execute their own tasks by themselves and work cooperatively to finish semantic query. KS3W intelligent system can not only break the traditional keyword-based search framework, but also generate illustrations reports for users. Following, an example is showed to illustrate how to use such a system to implement semantic search on the Web.Since the introduction of the Semantic Web, the Semantic Web searching technology has achieved fruitful results. More and more experts and developers have been involved in this area, and promoted the further development in this field. Semantic searching system involves a number of fields, such as, ontology, logical reasoning, natural language processing, machine learning, human-computer interaction interface, information retrieval and so on. There is no doubt that their respective progress and development will promote the development of Semantic Web, and vice versa.