Catálogo de publicaciones - libros

Many communities provide Web resource directories to help users find useful resources in the community. A typical example is a resource directory in a homepage of a local government. Crosswalk of the directories of neighboring communities is a crucial function for users to collect useful resources from the communities. However, an appropriate scheme bridging the community directories is required. This paper proposes a few mapping schemes to connect community directories and compares them by applying them to the resource directories of three local governments - Tokyo and Hokkaido in Japan and Chungcheongnam-do in Korea. The mapping schemes use National Diet Library Subject Heading (NDLSH) and/or Nippon Decimal Classification (NDC) as a switching language. Evaluation of the proposed schemes shows their advantages and limitations.

Provides a brief overview of the activities of the IFLA Functional Requirements for Subject Authority Records (FRSAR) working group. Introduces the group’s terms of reference and the work completed so far, including definitions of user tasks and subject entities. Discusses the development of the entity-relationship conceptual model of subject entities in the bibliographic universe.

The partial success in inferring the characteristics of offenders from their criminal behaviour (‘offender profiling’) has relied on limited data and subjective judgments. We therefore sought to determine if Information Retrieval techniques and in particular Language Modelling could be applied directly to existing police digital records of criminal events to identify significant characteristics of offenders. The categories selected were gender and age group. Results showed that distinct differences in characteristics do exist.

Most gazetteers with a global scope contain few local names, and gazetteers with a local scope mostly do not contain foreign names. However, people often use both local and foreign place names in a discourse. We describe some of the challenges in mapping local and global gazetteers that serve different needs and hence may have different structure, granularity and coverage. We pay special attention to the problem of identifying duplicate place descriptions in such registries.

This research aims to develop a model for identifying predictive variables for the selection of the best quality answer in a question-answering (QA) system. It was found that accuracy, completeness and relevance are strong predictors of the quality of the answer.

In this paper a two dimensional two directional pairwise Fisher’s linear discriminant (FLD), (2D pairwise FLD) is proposed which is employed for representation and recognition of Kannada numerals. The proposed methodology is robust as it is an extension of 2D pairwise-FLD[3] which is theoretically more efficient than conventional FLD.

At present, many researches of noun phrase translation are proposed in Natural Language Processing field, but most of them are in dictionary-based with word-by-word translation and similarity selection. The paper proposes a hybrid approach for noun phrase translation, by combining the set theory and grammar’s pattern, and its algorithms, to apply to Vietnamese–English translation. The finding also has good experimental results when applying on Vietnamese noun phrases.

This paper presents a new algorithm called “tree-structured networks” that can automatically construct parent-child (hierarchical structure) and sibling relationships (non-hierarchical structure) among concepts from a set of documents without use of data reduction or standard clustering techniques. The algorithm is applied to bibliographic databases such as INSPEC and EI Compendex toward the objective of enhancing research and development (R&D) management. Deriving tree-structured networks of research topics is an important goal in R&D management study. Parent-child relationships can help identify emerging areas in an existing field of research. Sibling relationships are interesting as well since they could represent interdisciplinary structures among related topical areas. Based on the initial testing on a set of publication abstracts, the proposed algorithm promises to offer richer structural information on relationships in text sources over the standard clustering techniques.

Social tagging systems are a popular means for sharing resources. However, social tagging depends on individual knowledge. We evaluate the effectiveness of tags in describing the resources using support vector machines via classification. We achieved precision and recall at 90.22% and 99.27% respectively, with an average accuracy of 89.84%. Our results show that tags may help users’ group resources into broad categories.

Digital Libraries currently use several advanced information technologies to organize information and make it easy accessible to users. Current digital library trends to be dynamic digital library [1]. It is possible that business rules also can be approached for improving dynamic digital library. Business rules [2] are statements that define or contain some aspects of IT systems by providing a foundation for understanding how an IT system functions. At present, the need for automated business rules is becoming more essential because of the increasing usage of IT systems. However, it is not easy to extract business rules because they are written in a natural language structure and much of it is ignored. Therefore, one important question in this research area is how to automatically extract a business rule from a document? Based on this, information extraction (IE) [3] typically can be applied. Basically, IE is to transform text into information that is more readily analyzed. We believe that if the content of a document is decreased, the accuracy of rules extraction may be increased logically. With this assumption, if irrelevant information is filtered from the document, it is possible to easily extract business rules from the rest. Therefore, this research proposes a method based on probabilistic text classifier to extract synopsis information. It could be said that this work is the pre-processing of a business rules extraction methodology.