Norihide Shinagawa

CyLog/Crowd4U: a declarative platform for complex data-centric crowdsourcing

This demo presents a principled approach to the problems of data-centric human/machine computations with Crowd4U, a crowdsourcing platform equipped with a suite of tools for rapid development of crowdsourcing applications. Using the demo, we show that declarative database abstraction can be used as a powerful tool to design, implement, and analyze data-centric crowdsourcing applications. The power of Crowd4U comes from CyLog, a database abstraction that handles complex data-centric human/machine computations. CyLog is a Datalog-like language that incorporates a principled feedback system for humans at the language level so that the semantics of the computation not closed in machines can be defined based on the game theory. We believe that the demo clearly shows that database abstraction can be a promising basis for designing complex data-centric applications requiring human/machine computations.

X2QL: An eXtensible XML Query Language Supporting User-Defined Foreign Functions

With the recent and rapid advance of the Internet, management of structured documents such as XML documents and their databases has become more and more important. A number of query languages for XML documents have been proposed up to the present. Some of them enable tag-based powerful document structure manipulation. However, their contents processing capability is very limited. Here, the contents processing implies the similarity-based selection, ranking, summary generation, topic extraction, and so on, as well as simple string-based pattern matching. In this paper, we propose an extensible XML query language X2QL, which features inclusion of user-defined foreign functions to process document contents in the context of XML-QL-based document structure manipulation. This feature makes it possible to integrate application-oriented high-level contents processing facilities into querying documents. We also describe an implementation of an X2QL query processing systemon top of XSLT processors.

Dynamic Generation and Browsing of Virtual WWW Space Based on User Profiles

Due to the recent and rapid advance of the WWW technology, a huge amount of information is available via the Internet. However, it is difficult for users to find and utilize relevant information. Generally, querying and browsing are both indispensable to access to the required information. In this paper, we propose a scheme to support the browsing phase of the information search based on specification of the user’s interest. In our scheme, each user is given virtual WWW pages reflecting his/her interest. Each virtual WWW page is dynamically generated based on the logical document structure extracted from the original page. Our approach makes it possible to browse the WWW space from his/her viewpoint and to identify required information more easily. We also show how such a scheme can be realized in a non-intrusive way in the current WWW browsing environment.

Constructing XML views over binary data

Ordinary applications mainly use application-specific binary data and XML has been widely used to exchange, integrate and manage data. When we expand the scope of XML technologies to cover binary data, it becomes possible for the data to be shared by many areas beyond that of the original application. We propose a scheme to achieve XML views over binary data, which are produced dynamically and partially based on address calculation. A DOM API supporting updates and XPath query optimization scheme for the XML views are then possible prototype. This paper presents experimental results using the implementation.

Efficient evaluation of human-powered joins with crowdsourced join pre-filters

There are an increasing number of crowdsourcing systems that exploit the power of the crowd to solve a wide variety of problems that cannot be solved only by computers. The human-powered join is one of the core operations that appear in many crowdsourcing systems, and existing papers have proposed efficient processing schemes for human-powered joins. However, the proposed approaches rely on assumptions that do not necessarily hold in practice. This paper proposes the crowdsourced join pre-filter, a novel technique for efficiently processing human-powered joins. The paper also shows theoretical results and gives the results of a preliminary experiment that suggest that the technique is effective.

C-Mapping: a flexible XML-RDB mapping method based on functional and inclusion dependencies

So far, many XML-RDB mapping methods have been proposed, but all of them deal only with either one-to-one or one-to-N mapping from XML elements to relational attributes, even though other mappings, such as N-to-one mapping, are possible. To extend the expressiveness, this paper proposes C-Mapping, which enables us to deal with N-to-one mapping as well as one-to-one and one-to-N mappings in XML-RDB mapping. C-Mapping takes functional and inclusion dependencies in XML data as input, and generates an XML view constructed over relational tables. An important feature of C-Mapping is that, since dependencies in XML data are captured and represented in XML-RDB mapping, we can make the best use of RDBs functionality to maintain the consistencies in the XML view. In addition, we show that C-Mapping is powerful enough to simulate a variety of mapping schemes proposed so far.

Integration of WWW Applications Based on Extensible XML Query and Processing Languages

The rapid advance of the Internet has brought with it a variety of WWW applications. Integration of these WWW applications, which implies development of meta level applications (shortened to meta-applications), is an important research issue. In the Internet context, XML has attracted a great deal of attention and will be used as a standard data format in WWW applications. This paper proposes a scheme to develop meta-applications on top of XML-based WWW applications. Our approach uses an XML query language called X2QL to manipulate XML instances. X2QL features the inclusion of user-defined foreign functions so that we can define functions specific to WWW application integration, such as submitting forms and following hyperlinks. In addition, we propose X2PL, which is an extensible XML processing language. In X2PL, processing flows composed of embedded X2QL queries are specified. This paper also describes an integration example and the development of a prototype system.