Combined with NSF Post Digital Libraries Futures Workshop in June 2003, this paper analyzes the achievements and limitations of the U.S.A. digital libraries research and development in the past ten years. Moreover, this paper discusses research goals and directions of the post digital libraries age.
This paper introduces the experiences and practices of establishing the Tsinghua Univerisity theses and dissertations full-text datatbase, including the system development, data production, copyright protection, preservations and backup and so on. On this basis, the paper puts forward some suggestions about the establishments of university theses and dissertations full-text database.
This paper demonstrates the design thought of multidatabase retrieval system based on multi-agent using the principle of meta-search engineer. The interface agent, cooperation agent, gathering agent, search deputy agent are used to achieve the intelligent part in this system. Communications between agents in this system is achieved through a mobile agent, so they can coordinate with each other to complete search task. Also the key technologies in the system are studied and discussed in this paper.
This paper presents an algorithm of self-adaptive matching method in Chinese segmentation. This algorithm not only identifies Chinese words in vocabulary successfully but also identifies unlisted words which are not in vocabulary on basis of decided vocabulary automatically. The test which compares this algorithm with Reverse Maximum Matching Method and some methods which identify unlisted words proves that it can resolve unknown words segmentation effectively, decreases mistakes of Chinese segmentation and has no effect on the efficiency of Chinese segmentation largely.
A teaching and researching aid system platform based on semantic Web is developed. It is based on current China discipline system and uses Web Ontology Language OWL as the representing language. This method makes distinctions among different types of knowledge representation. An example is given to explain the implement of this representation method.
This paper analyzes the problems of semantic disambiguation of cross language information retrieval, proposes that multilingual ontologies can improve the translation efficiency of cross language information retrieval and discusses two systems on it. Then the paper designs an Ontology-based cross language information retrieval model and describes the methods to realize it.
In view of the existing limitation of search engines in the information retrieval, this paper proposes multi-agent intelligent retrieval system model based on Ontology, and gives the system structure, workflow, function description of the model. Intelligent Agent utilizes ontology knowledge to normalize retrieval request information, which can improve the accuracy rate and cover rate of retrieval. Agents collaborate divisibly to finish information retrieval and automatic update service, and embody the intellectualization and individuality of system,etc. This model can provide foundation for realizing the highly effective intelligent retrieval system research.
Kansei Engineering is a quite new research field. The paper introduces the background of Kansei Engineering, its research content, common methodology and applications. Then with the respect to the issues of image retrieval, the paper discusses how Kansei Engineering could be applied in establishing image retrieval system based on Kansei features. Finally, it points out the major problems awaiting solution both Kansei Engineering and image retrieval system based on Kansei features.
This paper demonstrate the method of how to map the primitive image features to the semantic interpretations of the image content, and how to implement interactive learning together with probabilistic search. Lastly, this paper introduces the concept of Ontology for the image retrieval.
This paper introduces a search engine, which is designed for personal user. By using heuristic real-time search algorithm, it can provide user with the newest topic-specific information. This system can meet user’s need, and solve the problem such as topic fixation and data outdating which are ubiquitous phenomena in the general search engine. At the same time it provides theoretic and practical basis for the personalization of search engine.
This paper presents a new clustering algorithm based on GA(Genetic Algorithm) and k-medoids algorithm. The new algorithm can not only improve the precision of clustering but also recognize isolated points. At the same time,the new algorithm may expedite the convergence of GA and save the time cost for integration with the kmedoids algorithm in GA.
Based on finite state automaton, a new finite state automaton, named Scheme Automaton is proposed in this paper. On the basis of the model, a new Chinese word automatic segmertation model is designed, and also gives the key data structure and construction algorithm. Then analyzes the complexity of the algorithm.
In the light of the problems in the present teaching resource system, such as the single information and coincident integration and so on, the functional structure between share and cooperation is described, which are supported by the teaching resource system under the developing teaching surroundings. The technology of software integration based on the Web service is stated. An integrated frame of teaching resource system based on the services grid is presented.
This paper aims to rebuild the index system of evaluating Internet resources by artificial neural networks. Indicators and their frequency are studied out primarily by searching related articles, Original and weighted grade tables are obtained by questionnaire, Statistica Neural Networks (SNN) and SPSS are used to deal with the data, and the first-degree target system of evaluating Internet resources is achieved successfully.
This paper first discusses the applications of Web log mining in library , then comes up with a method of building a Web log mining system based on SQL Server 2005 . The function and implement methods of main modules are introduced and also a framework is designed. At last, the advantages of this system are discussed.
This paper builds a framework of Web usage mining based on XML technology, introduces XGMML and LOGML briefly. Then, the author discusses the method to generate LOGML documents. At last, the Apriori algorithm was used to mine frequent sets, frequent sequences and frequent sub-graphic in Web usage documents.
Learning from the thinking of developing information system, partners selection can be divided into three phases, which are qualification selection phase,core selection phase and expectation selcetion phase.The paper also blueprints the architecture of partner evaluation and selection system based on P2P and discusses two models, partner resources finding model and visiting controller,which are important to build up the search and login function of system.
This paper indroduces the two main deadlock models in the distributeddatabases， and analyzes four distributed deadlock detection algorithms. Then it presents an ameliorative deadlock detection algorithm- creating the dynamic DDA. This algorithm,which absorbs the advantages of existed deadlock detection algorithms and avoids their disadvantages, can well adapt the need of distributed database systems.
The author designs the interactive voice response system (IVRs) based on finite state diagrams. IVRs provides a new service mode and improves the service level for library.
The book checking and accepting in library is a complex work. Taking arrived electric list provided by suppliers as checking resource, the author designs and develops a new model for book checking and accepting automatically, which makes book checking and accepting more automatically. This model saves not only the flow of correcting titles, writers and prices with manpower check and acceptance, but also the step of recording amount of arrived books.
In library catalogue, universal Chinese author number is always welcome by library managers and readers. The main reason is that the weave method is relatively scientific. As author number check number method that goes down to the present has strict weave rule, it is relatively numerous and inefficient. Aim at the problem, the paper brings forward a new method .the method has the work automatically achieved by computer, so that the weave work become simple, shortcut, accurate and higher efficient.
Based on the study of marine life classification, this paper proposes the system organization way using the theory and technology of modern information management, which take the marine life picture information processing as the focus,puts emphasis on the method of system standardization design, classified and sign note and the method of retrieval of marine life picture.
This paper studies the strategy of automatic extracting and assembling papers system, and put emphasis on the construction of test-question storeroom and the strategy of extracting and assembling subject. It the same time brings forward dynamic algorithms of configure assembling parameters and model of random assembling subject. This paper also describes the main function of test-question storeroom subsystem, extracting and assembling papers subsystem and compiling papers subsystem in detail, as well as the corresponding interface based on Virtual C++ 6.0.
