Analysis of multiple update techniques on a RDF keyword search system

Keyword search is a technology that allows non-expert users to explore and retrieve information and it is traditionally used for unstructured data, such as in Web page searches. In the last decade, this search method has also become popular for exploring structured data, such as relational databases or graphs. Instead of using complex SQL or SPARQL queries and when the underlying schema is known, the user writes a series of words(keywords) to search for what he or she needs, getting as answers the ones more matching with the search. Keyword search systems are challenged by two fundamental parameters, efficiency and effectiveness. In fact, efficiency and effectiveness are two qualities of a SPARQL, or SQL, query that returns an answer quickly and always accurate even when operating on large amounts of data. The "virtual documents" method allows keyword search systems to work also on large databases by generating answers to keyword queries in a reasonable time. This paper aims to replicate the keyword search systems based on "virtual documents" TSA+BM25 and TSA+VDP for RDF graphs. In addition, two methods of update processing in a keyword search system, will be presented and analyzed: BruteForce and semiTSA. Although keyword search is a growing research matter, the topic of updates on structured data, such as RDF data, had not yet been addressed in the literature.