=Paper=
{{Paper
|id=Vol-1625/abs2
|storemode=property
|title=None
|pdfUrl=https://ceur-ws.org/Vol-1625/abs2.pdf
|volume=Vol-1625
}}
==None==
<pdf width="1500px">https://ceur-ws.org/Vol-1625/abs2.pdf</pdf>
<pre>
Topic Modeling without Generative Probabilistic
   Model: An Approach and its Validation ?

                        Mikhail G. Kreines, Elena M. Kreines

         BaseTech LLC, Raketny boulevard, 11-7-164, Moscow, 129366, Russia
                                mkrf@yandex.ru



        Abstract. We propose a novel approach to compute a model of seman-
        tic structure of natural language texts corpora based on the models of
        the texts. This approach differs from LSI and standard ways of topic
        modeling (D. Blei, 2012). We do not use matrix factorization and prob-
        abilistic distributions of the words given a priory. As the base for the
        corpora model construction we use the set of the original vector models
        of the corpora texts. The ways of empirical validation and applications
        of the models are considered.

        Keywords: natural languages, texts, text collections, models, Informa-
        tion Retrieval




?
    The work is partially financially supported by the Ministry of Education and Science,
    Russian Federation (Contract N. 14.579.21.0090 from 27.11.2014, project identifier
    RFMEFI57914X0090).

</pre>