Creation of Necessary Technical and Expert- Analytical Conditions for Development of the Information System of Evaluating Open Text Information Sources’ Influence on Society

Rustam Mussabayev, Bek Kassymzhanov, Aidos Mukashev, Viktoriya Ibrayeva, Azat Merkebayev

2019

Closed access Metadata (CSL-JSON)
DOI 10.1109/opcs.2019.8880193
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

In this paper, we trained distributional models (patterns) for text preprocessing in Word2vec and Glove. Three variants of text preprocessing were used to train distributional patterns. Based on the implemented distribution model Word2Vec, a vector representation was obtained for a cluster-separated test sample of 30 news items. All variants of the weighted average calculation of the vector representation of texts were considered. Two-stage clustering was carried out. After training the Doc2Vec model on normalized documents, a vector representation was obtained for each document. The following news about the same event was selected for the test, but from different sources. A 2-dimensional “factual cube” was analyzed.