Nowadays, Wikipedia has become one of the most important tools for searching information. Since its long articles are taking time to read, as well as section titles are sometimes too short to capture comprehensive summarization, we aim at extracting informative phrases that readers can refer to. Existing work on topic labelling works effectively and performs well on document categorization, but inadequate for granularity of detailed contents. Besides, existing keyphrase construction methods just perform well on very short texts. So we try to extract phrases which represent the target section content well among other sections within the same Wikipedia article. We also incorporate related external articles to increase candidate phrases. Then we apply FP-growth to obtain frequently co-occurring word sets. After that, we apply improved features which characterize desired properties from different aspects. Then, we apply gradient descent on our ranking function to obtain reasonable weighting on the features. For evaluation, we combine Normalized Google Distance (NGD) and nDCG to measure semantic relatedness between generated phrases and hidden original section titles.

Entity linking (EL) is the task of mapping name mentions in web text to their entities in a knowledge base. Most of earlier EL work in the knowledge based approach is usually formulated as a ranking problem, either by (i) non-collective approaches with supervised models, or (ii) collective approaches by leveraging global topical coherence which means semantic relations between entities through graph-based approaches. For the mapping process, we can regard it as selecting an entity to its mention by combining these two methods. In this paper, we propose a probabilistic model that ranks related entities to name mentions where ranking is customized by using three types of data: popularity knowledge of the entity, context similarity between mentions and the entity, and semantic relations between mapping entities. Specifically, we first propose an EL model utilizing global topical coherence that means semantic relatedness between entities, as well as using local mention-to-entity compatibility, to improve recall and precision. The key benefit of our model comes from 1) combination of two methods to provide customized ranking for mentions, 2) the model can save a large amount of calculation by efficiently finding candidate combinations of entities through global semantic coherence.

The number of entities in large-scale knowledge bases has been growing in recent years. The key issue to entity linking using a knowledge base such as Wikipedia is entity disambiguation. The objective of our proposing system is to disambiguate entities in documents and link entity mentions to their corresponding Wikipedia articles. To this end, our system ranks the set of candidate entities based on relatedness by utilizing semantic features derived from Wikipedia category hierarchies and articles. In addition, to reflect contextual information of Wikipedia, we utilize word embedding for refining the ranking result of candidate entities. Our experimental results show that these features have given good correlation with human rankings in candidate relatedness ranking and the combination of features has high disambiguation accuracy on news articles.

As one of the popular social media that many people turn to in recent years, collaborative encyclopedia Wikipedia provides information in a more "Neutral Point of View" way than others. Towards this core principle, plenty of efforts have been put into collaborative contribution and editing. The trajectories of how such collaboration appears by revisions are valuable for group dynamics and social media research, which suggest that we should extract the underlying derivation relationships among revisions from chronologically-sorted revision history in a precise way. In this paper, we propose a revision graph extraction method based on supergram decomposition in the document collection of near-duplicates. The plain text of revisions would be measured by its frequency distribution of supergram, which is the variable-length token sequence that keeps the same through revisions. We show that this method can effectively perform the task than existing methods.

We present a new method of detecting privacy violations in the context of database publishing. Our method defines a published view V to preserve the privacy of a secret query Q if V and Q return no tuples in common, over all possible database instances. We then establish necessary and sufficient conditions that characterize when V preserves the privacy of Q in terms of the projected inequalities in the queries, both for conjunctive queries and queries with negation. We also show that integrity constraints have an effect on privacy, and derive a test for ensuring privacy preservation in the presence of FD constraints. The issue of privacy preservation in the presence of multiple views is investigated, and we show that it can reduced to the single view case for a suitably chosen view.

Integration of version and access control of XML documents has the benefit of regulating access to rapidly growing archives of XML documents. Versioned XML documents provide us with valuable information on dependencies between document nodes, but, at the same time, presenting the risk of undesirable data disclosure. In this article, we introduce the notion of relevancy-based access control, which realizes protection of versioned XML documents by various types of relevancy, such as version dependencies, schema similarities, and temporal proximity. We define a new path query language XVerPath over XML document versions, which can be utilized for specifying relevancy-based access-control policies. We also introduce the notion of relevancy class, for collectively and compactly specifying relevancy-based policies. Regarding efficient processing of access requests, we propose the packed version model, which realizes space-efficient difference-based archives of versioned XML documents and, at the same time, providing efficient evaluation of XVerPath queries. Experimental results show reasonable performance superiority over conventional methods, which do not utilize version differences.

Integration of version and access control of XML documents has the benefit of regulating access to rapidly growing archives of XML documents. Versioned XML documents provide us with valuable informations on dependencies between document nodes, but at the same time presenting the risk of undesirable data disclosure. In this paper we introduce the notion of relevancy-based access control, which realizes protection of versioned XML documents by various types of relevancy, such as version dependencies, schema similarities and temporal proximity. We define a new path query language XVerPath over XML document versions, which can be utilized for specifying relevancy-based access control policies. We also introduce the notion of relevancy class, for collectively and compactly specifying relevancy-based policies.

Access control policies of XML documents are often specified based on user roles and data content of the documents. Content-based authorization is crucial for providing fine-grained access control to data in XML document. Since authorization rules (authorizations, for short) use path expressions of XPath for locating data in documents, authorization definition is related to structure of the document. However, the structure of XML documents tends to change by various reasons such as application extension and information exchange between organizations. Therefore, authorizations must be revised whenever they become incompatible with a new structure of the document. As far as we know, no previous work has discussed the problem of transforming content-based authorizations for XML documents by using schema mapping information. We define classes for schema and document transformations that allow transforming authorizations without access to source and target XML documents. We propose an algorithm that computes authorizations of role-based access control (RBAC) model for a target DTD instance from given RBAC authorizations of a source DTD instance and schema mapping information under the specified classes of schema and document transformations while preserving the authorization policy of the source DTD instance.

Web based services and applications have increased the availability and accessibility of information. XML has recently emerged as an important standard in the area of information representation. XML documents can represent information at different levels of sensitivity. However, XML access control models proposed in the literature enforce access restrictions directly on the structure and content of an XML document. Therefore the authorizations, which specify access rights of users on information within an XML document, must be revised whenever the structure of the XML document is changed. We present two approaches that translate the authorizations for the transformed XML document. The first approach translates instance-level authorizations by using instance-level mapping between instance nodes of source document and those of transformed document. The second approach translates schema-level authorizations by using schema mapping between schema elements of the source schema and those of the target XML schema.

Binary decision diagram (BDD) is a data structure to manipulate Boolean functions and recognized as a powerful tool in the VLSI CAD area. We consider that compactness and efficient operations of BDDs can be utilized for storing temporary relations in bottom-up evaluation of logic queries. We show two methods of encoding relations into BDDs, called logarithmic encoding and linear encoding, define relational operations on BDDs and discuss optimizations in ordering BDD variables to construct memory and time efficient BDDs. Our experiments show that our BDD-based bottom-up evaluator has remarkable performance against traditional hash table-based methods for transitive closure queries on dense graphs.>

Efficient procedures for producing nested relations from networks are considered. Various types of nested relations are defined according to the interaction between the networks' and nested relations' schemas. Among these types, partitional normal form (PNF) and accordant nested relations are shown to be produced by an extended navigation, called touch information display (TID) navigation, which requires time proportional to the number of tuples in the relation. It is demonstrated that a network can effectively produce a variety of nested relations by navigation. The efficiency of production is classified according to the interaction between the schemas of networks and targets. Non-PNF and/or non-accordant targets are generated using duplicate depletion. On the other hand, PNF and accordant targets can be generated using only navigation, which requires linear time in the number of tuples in the relation. Furthermore, for the cases on non-PNF and non-accordant targets, the TID method exploits the existence of links, so that the size of the region of the subrelation upon which deletion of duplicated values should be performed is limited.>