EXAMPLE-BASED GRAPHICAL DATABASE QUERY LANGUAGES

Example-based query languages allow users to specify queries through a n example that is constructed graphically, thus utilizing the analogy between a semantically meaningful example and the query. These languag es provide a user-friendly graphical interface. Since the appearance o f the first example based query language (Query-by-Example) in 1977, m ore than a dozen such languages have been proposed and/or implemented. They are designed for different application domains (such as statisti cal and scientific applications, office automation, historical databas es, and spatial databases) and have different capabilities and express ive powers. Most example-based languages are based on revised versions of Codd's domain relational calculus. This survey compares the featur es of 12 example-based languages in terms of (1) query specification a nd interpretation, (2) object manipulation, (3) query language constru cts, and (4) query processing techniques. Only two human factors studi es in the literature have investigated user friendliness and performan ce of example-based query languages. They involve Query-by-Example and Summary-Table-by-Example. The first showed that a nonprogrammer can c ompose a query in Query-by-Example after only two or three hours of tr aining. The second, using nonnovice users as subjects and complex quer ies in the testing, showed that Summary-Table-by-Example queries take twice as long to formulate, although the proportion of essentially cor rect queries was 13 percent higher in Summary-Table-by-Example. Overal l, empirical evaluation showed that 58 percent of the Summary-by-Examp le errors were semantic, while this was the case in 67 percent of Quer y-by-Example errors. Thus, the experimental results showed some pluses and minuses for both languages.