This paper introduces the difference between the Article-Level Metrics (ALMs) and the traditional bibliometrics, and summarizes the current cases of ALMs applications. By using the case study method, the author draws-up the operational mechanism of PLoS-ALMs in which includes its data sources, its metadata and raw data, and its open data. The discussion concerns about how ALMs supports open innovation development in future.

In this paper, the Chinese Agricultural Thesaurus (CAT) is expressed formally and described in a semantically linked way based on SKOS and SKOS-XL. CAT is also mapped and linked to other well-known knowledge organization system such as AGROVOC, NALT, EUROVOC and LCSH at individual level. In addition, a linked data publishing system of CAT is developed based on the open source tool Virtuoso, providing services such as HTTP URI dereference, CAT concepts browsing and navigation, SPARQL query endpoint and RDF Dumps. This study can advance the CAT's visibility, accessibility and interoperability with other systems, and lay fundamental base to describe, organize and semantically link other agricultural information resources as well.

In order to specify and organize major resources in domestic and foreign research areas, the authors build the knowledge base based on scientific research Ontology. The purpose is to accomplish resources'collection, specification, organization, preservation and showing to support knowledge discovery. This paper develops the methods and related tools for data collection, data specification, and data integration, and then tests the effect by the practice of construction data, forming a certain scale of scientist network.At last, the perspectives of these methods are discussed about the sustainable building for knowledge base.

This article aims to summarize the quality of social tags. Firstly, the background of the tags quality assessment issues is expounded. Secondly, the existing methods of tags quality assessment are summarized and compared, and the problems of the tags quality assessment are pointed out. Finally the social tags quality evaluation studies and the future direction of development are put forward.

The paper summarizes the research of relation extraction techonology. It firstly gives a brief overview of relation extraction,and divides the research into two phases: the relation extraction in specific domains and the relation extraction of Web text. Then,analyzes the major methodologies of the two phases: the relation extraction in specific domains mainly uses machine learning methods with annotated corpora, while the relation extraction of Web text uses rule-based methods or distant supervision methods according to different demands.

This article uses SEM and regression analysis to empirically demonstrate relationships among internal control mechanisms, shared mental model, knowledge activity and knowledge sharing, test results reveal that knowledge activity plays mediating role in relationships among task shared mental model, collaboration shared mental model and knowledge sharing respectively. Behavior control plays a positive adjustment role in relationships among task shared mental model, collaboration shared mental model and knowledge activity respectively. Result control plays a negative adjustment role in relationships among task shared mental model, collaboration shared mental model and knowledge activity respectively.

Based on semantic relationship of CMeSH and characteristics of biomedical vocabulary words, a semantic auto-mapping from literature keyword to subject headings is presented, including semantic similarity calculation model between literature keyword and subject headings of CMeSH, and the best subject heading identification scheme. Experimental results of disease and microbes vocabulary are analyzed.

The processing of recognizing and extracting the reviews from webpages is transformed into recognizing the DOM-SubTrees which is cyclical in the DOM-Tree. Each node is iterated in the DOM, and the similarity between DOM-SubTrees is calculated, then those nodes meeting the requirements are found out.The proposed method can calculate the similarity between DOM-SubTrees in the end. To make it suitable in recognizing the reviews in webpages, the paper transforms the DOM-SubTrees into the paths of leave-nodes which consider the name and the position of tag.The authors compare 4 methods which are used in calculating the similarity between DOM-SubTrees, and also compare the algorithm with other algorithms which recognizes the reviews in webpages by using the weight of tags in the DOM-Tree. The experiments show that the algorithm has higher precision and recall rates, and more effective than other algorithms.

This paper puts forward a new co-word analysis process model according to the deficiency in tradition co-word analysis. This model improves the traditional methods of co-word analysis from two aspects. At first, this paper supplements the indexing keywords because they cannot fully describe the topic content of the thesis. High frequency words from indexing key words are chosen to constitute a supplementary dictionary. Paper candidate keywords are extracted from the title by the word segmentation technology based on the supplement dictionary, and then the candidate keywords are supplemented according to certain rules. Secondly,domain Ontology is introduced to calculate the high frequency keywords for semantic similarity because the co-occurrence frequencies are difficult to accurately describe the similarity between two words,considering the co-occurrence frequency and semantic similarity. Then the correlation is used to describe the word similarity, and is the basis of building co-word matrix. Finally, experiments prove the effectiveness of this improved method.

In order to study the features of microblog following relationship network, the analysis method based on complex network is applied to analyze the following relationship of Sina Microblog in this paper.Firstly,the K-core decomposition operation is performed on a microblog following relationship network to obtain a core user's relationship network. Secondly, the features of online community network and those of realistic society network can be received in the K-core network by computing the basic parameters of the K-core network, following ratio and degree correlation. Finally, a conclusion can be obtained that the community characteristic of the network structure is very obvious by community detecting on the K-core network and analysis of the node centrality, reciprocity and the role of brokerage. Experimetnal results show that the research work of this paper can provide effectively a fundamental empirical analysis for related applications.

This paper proposes a new approach to detect internet deceptive opinion based on Support Vector Machine(SVM). According to the different features in different opinions or sentiments, the authors firstly evaluate the effect or importance of various features. Then a deception detection model is developed with SVM. This model adopts polynomial kernel function and RBF kernel function after optimization to generate the classifier. The results of the experiment show that the proposed method is effective in identifying the deceptive opinion.

This paper analyzes the two defects of traditional presentation methods for retrieval results, and proposes a method of information retrieval results visualization based on social network. Starting from the medium dimension of scientific research authors and with the purpose of figuring out the relationships among the document authors, the paper builds an adjacent matrix based on the authors' research network structure and explores the invisible knowledge retrieval problem under the academic information aggregation and association. The re-aggregation of the retrieval results in form of a graph are directly shown to the users in order to achieve the purpose of improving digital library knowledge service.

According to the problem that in complex network environments, electronic resources of library sometimes fail to be browsed or downloaded, this paper demonstrates a program, by means of operating webpage document objects and so on, to automatically examine and feedback the availability of electronic resources. This program can monitor the status of Web server and check whether the end users can browse or download the resources, and thus the administrator can maintain the server in time to improve the utility of resources.

The high cost of information push and large data flow are the main problems in the current mobile phone library information push systems. This paper introduces a proposal to design a new library information push system based on GCM. The requirements analysis and technical ideas, the mobile client implementation, the information providing server side implementation and the service operating results are elaborated. The system implements a low cost, low flow push library information in real time and meets readers' demand for personalized information subscription.