
May 30 - June 3, the largest Russian computer-linguistic conference Dialog will be held in the Moscow-based boarding house Bekasovo. In detail about what “Dialogue” is and why ABBYY is organizing this conference, we have written in detail
here .
This year the main topics will be:
Estimation of text (sentiment analysis) . To solve this problem (how to understand the author’s attitude to what he describes) are used both methods based on linguistic rules and computer-aided training methods on large test collections of documents (in which experts manually set tonality assessments, and the computer tries to figure out which It is the properties of the test text that are associated with the evaluation in order to evaluate new texts on their basis). I think many have come across “correct” assessments of the tonality of articles in the Russian media monitoring systems (we will not give names), so the topic is very relevant.
')
Creation of new texts for linguistic studies . What is a shell? The corpus of texts (this may be speech recordings) is the research material on which computational linguistics tries to build models for automatic language processing. Modern enclosures can include millions of specially selected and processed texts. But even such cases have not enough for the powerful algorithms of statistical processing, which are used today. Therefore, the task is to create such cases automatically using the Internet as an almost unlimited source.
The tradition of the “Dialogue” is to hold
competitions of automatic document analysis systems . The goal of such competitions is not sports, but research: the development of reliable criteria and methods for evaluating automatic analysis systems. This year, under the auspices of the “Dialogue”, two such competitions were held: testing the systems for parsing texts in Russian (parsers) and testing systems that evaluate the tonality of texts in Russian. The results of these tests will be summarized in the Dialogue (after the end of the conference we will publish them in this blog, stay tuned).
Since Dialogue is an international conference, it is traditionally attended by world-class computer linguistics experts. This year,
Dan I. Moldovan is a professor of computer science at the University of Texas at Dallas, USA, and
John A. Carroll is a professor of computer linguistics at the University of Sussex, United Kingdom.
Conference working languages ​​are Russian and English.
Applications for participation are accepted until May 28 at
secretary@dialog-21.ru . All additional information you can find
on the conference site "Dialogue" .