Charles Welty

Human factors comparison of a procedural and a nonprocedural query language

Two experiments testing the ability of subjects to write queries in two different query languages were run. The two languages, SQL and TABLET, differ primarily in their procedurality; both languages use the relational data model, and their Halstead levels are similar. Constructs in the languages which do not affect their procedurality are identical. The two languages were learned by the experimental subjects almost exclusively from manuals presenting the same examples and problems ordered identically for both languages. The results of the experiments show that subjects using the more procedural language wrote difficult queries better than subjects using the less procedural language. The results of the experiments are also used to compare corresponding constructs in the two languages and to recommend improvements for these constructs.

Correcting User Errors in SQL.

Abstract Previous human factors research on SQL has discovered many correctable errors made by users. An experiment was run, testing how seriously error correction would affect SQL user performance. In the study, 39 subjects used SQL without error correction and 40 subjects had specific categories of errors corrected. The main result was that error correction improved user performance by 26%.

Experimentation in computer science: an empirical view

In many disciplines, scientific inquiry relies heavily on experimentation. Computer science is compared to other scientific disciplines in its use of experimentation by classifying articles in professional journals as experimental or non-experimental. The results of the classification suggest that experiments occur less frequently in computer science than in many other disciplines.

Usage of and satisfaction with online help vs. search engines for aid in software use

Computer users have long been frustrated by software problems. It is unusual that the Help menu actually helps with the software problems they have. At the same time, computer science students and professionals have been using search engines to get help with the complex software they use. The use of search engines to get help with software by both computer scientists and students in other disciplines is investigated. Students from all disciplines tested were found to use and be more satisfied by search engines than Help. Further investigation showed that, generally, students went to other people more than Help or search engines but found search engines and people to be the most satisfactory sources of help. Recommendations are made to improve Help systems by incorporating aspects of search engines.

Natural language querying (session overview)

The interface between a user and a software system should match the needs and background of the user. Many query languages are built to aid clerks and managers in the performance of their jobs. Therefore, a good query language would require no low level programming skills of its users. Usually very high level nonprocedural languages are used because they eliminate low level artifacts such as loops. An appealing high level language would be the users natural language. Natural language should require no training or refresher courses for successful use. Unfortunately, natural language interfaces have their own problems. Ambiguity is a major problem and can arise in a variety of ways. Ambiguous modifiers (e.g., “List the growers of orange trees.”) and ambiguous pronoun references (“Who was Doug Fluties high school coach? Also, what are his statistics?”) are just two examples. A casual user of the system may also assume a shared background with the system. The Doug Flutie query assumes the system knows that Flutie is a football player and that his (or his coachs) statistics refer to football achievement; not academic, physical or other measures. Resolving these and other problems requires an extremely complex and, as yet, unachieved interactive interface. This complexity can be reduced if we settle for something less than unrestricted natural language. One simplification is to restrict the domain of discourse to the specific database being queried. Other possible restrictions would be on the allowable form of a query, vocabulary, pronoun references, etc. Many questions arise about restricted natural language query systems. Some of the possible questions are:How do various types of restrictions affect use of a natural language query system? How can we combine the strengths of restricted natural language, menus and formal language in an integrated query system? How does a restricted natural language query system perform in the “real world” compared to a formal language query system? The three papers in this session address themselves to these questions, respectively.