Web Storage Interface

This article describes a draft and a implementation of Semantic Web data store one of the main parts of the Infrastructure for Semantic web. The main goal of this project is to create a new simple interface between any Data Store and other pieces of the infrastructure. This interface should not dependent on any existing or future Data store. It allows load data and execute query on them. The main part of this article will include interface for data import. The implementation takes advantage of a relational database Oracle and everyone can use this implementation from more than one programming language. Performance tests show not only bottle-necks of this solution but also that Data store has excellent load characteristic. These tests also determined improvements and future research. Introduction Everyone can imagine that the world wide web includes the great amount of information. In case of the Semantic web is situation very similar. In both types of these webs if a user wants to look up something specific he usually use one or more search engines. Every search engine is based at least on data importing into data store, data indexing and query evaluating. For storing and searching in Semantic data are developed some data stores. All of them are based on one of these principles: • store data in memory, • store data in native format or • store data in relational database. Each of these principles has an advantages and disadvantages. In-memory data store is very fast but it has limited capacity. Data store based on native format may be fast but each change in data store structure may be difficult. Data store based on relation database is slower than in-memory data store but it can store a huge amount of data. The most common data stores are Jena and Sesame in semantic web community. These data stores work perfectly with a small amount of data but when we try to work with a huge data the work is impracticable. So we try to design, implement and test a new data store for semantic web data. Each two existing data store interfaces are different so it is too difficult to change data store in semantic web application. There is no standard for data store interface. We propose a new interface that should become an universal interface for all semantic data stores. Pilot implementation do not use existing data store (e.g. Jena, Sesame) yet but use a new designed and implemented data store. In the future work we try to make implementation that connect the proposed interface with common used repositories. Infrastructure for Semantic web In [IFR2006], there was introduced proposal of Infrastructure for Semantic web. We consider that the main part of this infrastructure is a place, where anyone can save own data. This place can call Data Store. Many other parts of this infrastructure can interact with Data Store. Each of these parts have to specific function and each of them (see Figure 1) are bound to access to data, possibly insert new data. WDS'07 Proceedings of Contributed Papers, Part I, 110–115, 2007. ISBN 978-80-7378-023-4 © MATFYZPRESS