Automatic Knowledge Graph Generation For Cultural Relics From Web Text

With the development of the Internet technology, more intelligent search engines will be proposed. It becomes important to solve these problems of knowledge organization, storage and retrieval. Many researchers construct knowledge graph, which to reflect the relations between domain knowledge attributes in order to provide knowledge for the intelligent search engines. A antiques information knowledge graph is automatically constructed in this paper. The specific contents are as follows:(1) A algorithm of antiques information extraction based on rules is proposed. Firstly, identify possible antiques informations. Secondly, context words of possible historical relic match up features words to generate initial rule. Finally, generalize initial rule. This proposed method can reduces human intervention, in rules generation process. Further more, the proposed method increase extraction efficiency, under the same accuracy.(2) A algorithm of antiques information knowledge extraction is proposed based on Extreme Learning Machine(ELM) combining with rules. Firstly, take advantage of rules to extract Knowledge of historical relics with strong regularity, Then, historical relics not being taken out of the using ELM algorithm to extract. The experimental results show that F values were improved 7.22% and 4.01%, compared with only based rules and classification method. Further more, the training speed increased by 2 times, compared with Conditioal Random Fields (CRF) and Support Vector Machine (SVM).(3) A method of antiques information knowledge storage based on joint index of Resource Description Framework (RDF) is proposed. Making full use of local distribution and hierarchical morphological feature, the joint index of RDF achieves the rapid retrieval of antiques information. Experimental results show that this method proposed in this paper retrieval speed increased by 1.55 times, compared with the index based on six RDF storage, and retrieval speed increased by 1.42times compared with three retrieval based on three indexes RDF storage.(4) A antiques information knowledge graph generation method is proposed in this paper. Through continuous search with the same attribute value of historical relics to extend the knowledge graph, and in search of the same attribute values of historical relics in the process makes full use of the four attribute index, the algorithm has strong robustness and expansibility, reducing the manual intervention, it is of great significance to generate knowledge graph of other fields. In this paper, the author tries to build a historical knowledge graph and combined with the traditional search engine to analyse the application of information retrieval, finally found that based on knowledge graph of historical relics make retrieval results more rich and structured.Based on the above algorithmic process to achieve a knowledge graph historical relic generated automatically program and verify the effectiveness of the program.