A knowledge component extraction technology using figures and tables

Abstract

With the growing complexity of document contents and the significant increase of domain knowledge, it is difficult for knowledge receivers to understand specific domain knowledge. However, traditional knowledge extraction schemes usually provide complete documents to knowledge receivers and much time is required for knowledge receivers to acquire domain knowledge. The concept of component-based knowledge is to divide documents into several knowledge components corresponding to more specific domains, which can be used to reduce the time required for the knowledge receivers to search the specific domain knowledge. Moreover, since the figures and tables in a document usually contain the important implicit knowledge expressed within the document, the aim of this research is to extract the knowledge components from documents (e.g. industry yearbooks) on the basis of figures and tables. In this research, a knowledge component extraction model with two algorithms, namely the keyword mapping algorithm and sentence mapping algorithm, is developed. In order to demonstrate the applicability of the proposed methodology, a web-based knowledge component extraction system is also established based on the proposed model. Furthermore, Taiwan Logistics Yearbooks are applied as examples to evaluate the proposed model. The verification results show that the developed system is a high-performance knowledge component extraction system. As a whole, this research provides an approach for knowledge receivers to efficiently and accurately acquire domain knowledge. domain knowledge and, in doing so, they spent much time reading irrelevant content (Hou and Lin 2004).If documents can be divided into several knowledge components, each corresponding to more specific domains that are underpinned by component-based knowledge, then the reader can efficiently search for specific domain knowledge. As the reader digests similar knowledge components, in order to acquire specific domain knowledge, he or she can search for domain-specific knowledge quickly and accurately by properly focusing on specific informational components.Critically implicit knowledge from documents is often given in the form of figures and tables, and domain-specific knowledge is embedded within the paragraphs surrounding these figures and tables. Notably, these paragraphs, which hold important information or knowledge, can be efficiently digested using component-based knowledge. If knowledge components can be extracted from documents based on figures and tables, then the time required for the reader to search for specific domain knowledge can be effectively reduced.Previously, when readers searched for specific domain knowledge through traditional knowledge extraction schemes, they had to digest a great deal of irrelevant content from entire documents in order to obtain the necessary domain-specific knowledge. In traditional knowledge extraction schemes, figures and tables served as an auxiliary tool which provided the intended contents for the reader. As such, the time require...

Tagging documents using neural networks based on local word features

“…Another approach, often related closely with Natural Language Processing (NLP) techniques, is through the use of carefully crafted rules for keyword extraction, in the form of expert systems [12][13] [14][15] [16]. This approach may be effective in specialized domains, and closed archives, but they typically would require some manual construction of a set of inference rules on very particular linguistic styles and domains of knowledge.…”