A new piece of software to facilitate the management of surveys of CIS has been launched
Researchers at the Facultad de Informática UPM have developed a new piece of software for the Centro de Investigaciones Sociológicas of Spain which will save four years of work.The software will manage the survey database.
The Centro de Investigaciones Sociológicas is an organization under the authority of the Ministry for the Presidency. It was founded 50 years ago and since then it has been collecting data on the opinions of Spanish society. The centre has so far generated over 87,221 questions and their collected corresponding answers.
To conduct new surveys, the CIS does not need to ask new questions, what it needed is to use the numerous accumulated questions to update them and figure out the current opinion about these questions.
New surveys conducted by the CIS do not rquire new questions, Instead the program draw on the extensive database of questions of CIS, update them and then research current opinions using the questions
Homogenizing questions
In order to use the record of questions, the CIS has standardized the structure of the questions, which means to entitle the surveys to be able to organize and use them again for new surveys and to generate temporal series of questions.
The CIS used to have two people to do the job of classifying questions and to standardize the titles of surveys, they used to manually put together 39,257 questions for certain titles..
This piece of software has been developed in two months by the group of Validation and Industrial Applications of the Facultad de Informática, led by the Professor Jesús Cardeñosa. They achieved to group 22,347 questions, almost half of the needed. They classified the surveys with their respective titles. Thus, the program has saved four year of human labor..
This program has generated 96% of correct titles, according to the technical evaluation carried out on the legibility and relevant information in the questions analyzed.
The results of this research were presented by the Professor Carolina Gallardo at the lecture of Flexible Query Answering Systems, held recently in Belgium, and have been published in the proceedings of the lecture.
Reference:
Carolina Gallardo Pérez y Jesús Cardeñosa. Knowledge Extraction for Question Titling. Flexible Query Answering Systems, LNAI 7022, 9th Internacional Conference EQAS 2011, Ghent, Belgium, October 26-28, 2011, Proceedings, páginas 119 y ss.
Source: FIUPM