NLIDB Home Description Corpus Demo Bibliography
 

 

Corpora of Test Queries

 

There exist several versions of the NLIDB, each one having different capabilities. Version 1 includes three capabilities: solving basic query translation problems, dealing with the problem of semantically implied entities (which allows translating complex queries), and facilitating NLIDB configuration by using a wizard. Therefore, for version 1 there are three query corpora, one for testing each capability. This web page includes links to the three query corpora. 

The first corpus presents a set of queries for testing the effectiveness of the NLIDB for solving basic query translation problems. The corpus contains a set of 70 queries for the ATIS database (available at https://catalog.ldc.upenn.edu/docs/LDC93S4B/trn_prmp.html), which is a representative sample of the entire ATIS corpus.

 The perceptive reader might notice that many of the queries in the first corpus request information on flights that depart from one airport and arrive in another; while the queries in the original ATIS corpus request information on flights for traveling from one city to another. These queries are very complex because they involve the problem of semantically implied entities. The second corpus includes queries for testing the ability of the NLIDB for dealing with this kind of complex queries.

 The third corpus presents a set of exercises for testing the configuration wizard, which greatly facilitates fine- tuning the configuration of the NLIDB for querying different databases.

  1. Corpus for testing the effectiveness for solving basic translation problems.

  2. Corpus for testing the ability for solving complex queries (problem of semantically implied entities).

  3. Corpus for testing the easiness for fine-tuning the NLIDB configuration using the wizard.

 

 

 

Powered by Instituto Tecnológico de Ciudad Madero