Patents

   Custom applications related to Information Mining

   Voice Technologies

   Data Base Access in Natural Language

   Electronic Dictionary Based Applications

Patents

Jesús Martínez Barbero, Eugenio Santos Menendez. Patent P200802356, “Transmission Method for Multimedia Content” submitted on August, 8th 2008 to the Spanish Patent and Trademark Office and granted on June, 6th 2009. Submitted and approved the international PCT extension with application number PCT/ES2009/000382 and presentation date 20/07/2009.
Principio de la página

Custom Applications related to Information Mining

  • In the framework of the European Commission Project FLEX-Esprit IV nº 29158, we have developed applications that can be run in any environment. One of them is a data-mining system for authors of journalistic papers. The set of documents can be written in any format (Word, HTML, etc.) and can consist in a set of documents in data base or simply of a collection of documents. The system works as efficiently in English as in Spanish, and the same technology can address the data-mining of pieces of relevant information or, under request, of the author or any other type of information. This type of technology is specially interesting as it increases substantially the performance of search engines whether in data bases or the Internet, even more when it is made to measure for a set of documents. The more outstanding until now in the real tests made is the rapidity and accuracy. (97% of success).

  • Our work team has put a special interest on the technology addressing the determination of the subject of a document. It is frequent that when someone is looking for information on a subject in document data base or in the Internet, he/she is asked to define a set of key words that represents the domain from which he/she wants to retrieve information. It is obvious that the results of this kind of search are very limited. The technologies developed by our work team start with the text or paper that constitutes the information and automatically extract from it the set of key words that define the domain or subject of the paper. These key words can be compared with the request made by the user and can determine the documents that correspond to that request. Once determined this first document, the system can search the set of documents related to this subject or at least whose contents are close to it.
The application are numerous, as they go from more an efficient search of documentation to automatic readdressing of information that arrives in an organization to the corresponding departments, according the subject dealt with. This application is specially interesting for readdressing electronic letters sent to corporate addresses. Even more, this type of techniques can, observing the information requests made by an user during a period of time, determine his taste and interest profile, which allows, among other things, to offer him directly information made to measure and which supposes a better exploitation of the contents of an information provider, o even offer him directly a set of programs associated to his audience profile, without him having to define a list of favorites.
  • Using these techniques in a deeper way, you can make automatic summaries, above all of journalistic papers, and even generate from a paper, a headline. In specific domains, these techniques offer good precision, for example, for the elaboration of notes related to changes in the executive managers of a company. The generation of summaries in open environments requires a much more significant degree of technology. The performance is clear. You start with an input text, whatever long it is, and you end with a summary of it. We can say that, more than an application, it is a tool that would help people that have to do this job, tool whose technology, as we have commented before, is much more complex than previous technologies.

  • We have also developed an orthographic corrector of Spanish that can be adapted to any application. The same module can be extended in custom applications to validate the correction of data introduced as text in any other application. Principio de la página

Voice Technologies

Our work group has developed language modules that permit to handle any application through commands and that go from the use of computer applications (Internet Browsers o other applications) to the access to the information of a WAP data base without having to go through the information tree. It is a big problem as the user usually does not know the structure of the tree and often has to go through the whole tree before finding the information he needs, and even worst, nothing indicates if simply the information is not there. The application that permit to handle a computer application by voice, augment considerably the rapidity of use as the user does not have to move his hands from the keyboard to the mouse, they are particularly interesting for persons having motor handicaps, and have difficulties to use a computer application or, even for simple things as to change the TV channel, or using any type of device (elderly people, etc.)

Data Base Access in Natural Language

This application was developed some years ago in collaboration with the Company Software AG, that have a great experience in developing language applications. This technology can be customized and be supported by this company and our work group.

Electronic Dictionary Based Applications

Any electronic dictionary application, even for the development of Natural Language Systems, is important in intelligent interface applications. Our work group has developed during the past few years, an electronic dictionary of more than 30.000 Spanish entries, each of them associated with more than 30 information attributes. The applications are numerous and go from automatic translation systems to natural language processing systems. Combining the above-mentioned techniques of information retrieval with dictionary maintenance tools, we can enrich the dictionary with terms as we retrieve them in documents.
  • Besides the systems mentioned before, we have developed a lexical analyzer of Spanish that covers almost the totality of our language, and we are making a syntactic analyzer that permits, thanks to heuristics, to reach the semantic representation of a sentence. The applications go from the teaching of our language to, in a near future, a style corrector, which is very important if we want to achieve multilingual translation systems.

We can offer demonstrations to the persons interested in applying these technologies to their products, and in any case, you can contact us for joint or customized development. Principio de la página