6 .User Study Results As query complexity increased, autocorrection tool performs significantly than DM. (For the two easier task, DM was faster.) Users found autocorrection and display of non-answers to be useful for query debugging. Increasing Difficulty
7 .Auto-correction When query complexity increases, debugging becomes hard → autocorrection correcting misplaced tuples into answer and non-answer categories System looks at the space of all “tweaks” to the current query tree and recommends query tree
8 .Problem Existing query specifications makes it hard to specify quantified queries. VIQING: Visual Interactive QueryING, Olsten et al. (1998) Faceted Search
9 .Summary DataPlay: easier trial-and-error specification quantified queries. Nested universal table data model Syntax locality: Toggling of existential and universal quantifiers. “Tweaks” to queries could be done through graphical manipulation of query tree. Lack of non-answers: Query correction via examination and selection of answers and non-answers examples.
10 .Data Model nested universal relation=nested data model+ universal relations Benefits: no need for joins or group construction closed world assumption ∴ non answers can be easily computed as (universe - answers) The nested universal relation data model, Levene and Loizou (1994)