ecsentithemelex 0.0.1

Russian economic tonal-thematic dictionary

Russian tonal-thematic dictionary, which allows identifying the semantic orientation of groups of economic texts, as well as determining their sentimental (tonal) characteristics.

Table of Contents

1. Installation

2. Example of application

3. Authors

Installation

• Install the 64bit version of Python 3, for instance from https://www.python.org. Then run:

pip install ecsentithemelex

or

pip3 install ecsentithemelex

• Github:

https://github.com/ilya013/ecsentithemelex

Requirements: * NumPy (https://numpy.org/install/) * Pandas (https://pandas.pydata.org/getting_started.html) * pymystem3 (https://github.com/nlpub/pymystem3) * NLTK (https://www.nltk.org/install.html)

Example of application

First you need to import the library into the Python programming environment:

import ecsentithemelex

*ecsentithemelex* has three main modules: Vocabulary, OneWordScore and TextScore. The first, Vocabulary, contains tables with words that were collected and evaluated by the authors listed below and various experts, and these words were assigned to 12 different economic topics (in Russian):

array(['Международная экономика, торговля и финансы',
       'Государственные финансы, бюджет и налоги',
       'Реальный сектор экономики (производство, промышленность, добыча)',
       'Инновации', 'Общеэкономическое', 'Маркетинг и реклама',
       'Социальная ответственность, благотворительность, спонсорская деятельность, экология',
       'Монетарная политика, валюта, деньги и кредит, банки',
       'Потребление и розничная торговля', 'Фондовые и товарные рынки',
       'Корпоративные финансы и управление, фирма, бухгалтерский учет, нематериальные активы',
       'Макроэкономика'], dtype=object)

*Vocabulary* has three methods: all_phrase_tone(), all_word_tone_theme() and voc.all_forms_tone_theme(). All of them return DataFrame with word, scores or scores and topics. all_phrase_tone() returns all words, bigrams and trigrams with their scores. all_word_tone_theme() returns all words, bigrams and trigrams with their scores and categories. all_word_tone_theme() returns all words, bigrams and trigrams in all declensions with their scores and categories.

from ecsentithemelex import Vocabulary
voc = Vocabulary()
voc.all_phrase_tone()
voc.all_word_tone_theme()
voc.all_forms_tone_theme()

*OneWordScore* and *TextScore* use str object as input. For example(https://raw.githubusercontent.com/isdemin/ecs/master/news.txt):

word = 'фьючерсы'
text = 'Мировые цены на нефть перешли к росту, поднимаются на 1,5-1,7% в пятницу вечером после падения днем на 2%, рынки продолжают оценивать перспективы по балансу спроса и предложения, свидетельствуют данные торгов. По состоянию на 20.31 мск цена сентябрьских фьючерсов на североморскую нефтяную смесь марки Brent росла на 1,58% — до 43,02 доллара за баррель. Августовские фьючерсы на нефть марки WTI дорожали на 1,72% — до 40,3 доллара за баррель. Утром в пятницу снижение цен на нефть составляло 1%, днем достигало 2-2,5%. Трейдеры оценивают перспективы спроса и предложения после новостей от производителей нефти. Ранее в пятницу Международное энергетическое агентство (МЭА) в своем июльском докладе сообщило, что ожидает спрос на нефть по итогам 2020 года на уровне 92,1 миллиона баррелей в сутки, на 400 тысяч выше предыдущего прогноза.'

*OneWordScore* score and categorize only one word with methods score() and cateorize():

from ecsentithemelex import OneWordScore
ows = OneWordScore()

ows.score(word)
0

ows.categorize(word)
'Фондовые и товарные рынки'

*TextScore* score and categorize different texts without the need for their processing and lemmatization with methods score_text() and categorize_text():

from ecsentithemelex import TextScore
ts = TextScore()

ts.score_text(text, bigrams_in=True, trigrams_in=True)
0.011627906976744186

ts.categorize_text(text, bigrams_in=True, trigrams_in=True)
'Реальный сектор экономики (производство, промышленность, добыча)'

Authors

Below are the people who were directly involved in creating the dictionary, evaluating words, and searching for thematic categories for each word:

• *Fedorova E.A.*, prof. Department of Corporate Finance and Corporate Governance, Financial University under the Government of the Russian Federation, [e-mail]

• *Afanasyev D.O.*, JSC “Greenatom”, Moscow [e-mail]

• *Remesnik A.B.*, HSE, Faculty of Economic Sciences, [e-mail]

• *Demin I.S.*, prof. Department of Data Analysis and Machine Learning Financial University under the Government of the Russian Federation, [e-mail]

• *Sokolov A.V.*, HSE, Faculty of Economic Sciences [e-mail]

• *Pyltsin I.V.*, Higher School of Economics, Faculty of Economic Sciences, [e-mail]

• *Nersesyan R.G.*, LLC “Tcifra”, [e-mail]

• *Lazarev A.M.*, Lomonosov Moscow State University, Faculty of Mechanics and Mathematics, [e-mail]

• *Rogov O.Yu.*, Skolkovo Institute of Science and Technology, NS, Ph.D. of sciences [e-mail]

The power of the dictionary in this module may differ because the dictionary was supplemented with different word forms using the pymorphy 2 tools(https://pymorphy2.readthedocs.io/en/latest/) and pyphrasy(https://github.com/summerisgone/pyphrasy). This allows you to score and categorize texts without the lemmatization procedure, just tokenize the text and bring it to lowercase.

The algorithms of scoring and categorizing will be improved in next releases.

*Developed by Ilya Pyltsin*