Sentiment analysis support

[akolonin]

Need to provide sentiment analysis support for English, Russian and Chinese

Sources:

English:

1. https://github.com/aesuli/SentiWordNet https://raw.githubusercontent.com/aesuli/SentiWordNet/master/data/SentiWordNet_3.0.0.txt https://creativecommons.org/licenses/by-sa/4.0/
2. https://www.cs.uic.edu/~liub/FBS/sentiment-analysis.html#lexicon [Bing Liu]
3. https://www.kaggle.com/c/tweet-sentiment-extraction/overview/description

Russian:

1. RuSentiLex: https://www.labinform.ru/pub/rusentilex/index.htm
2. Linis-Crowd: http://www.linis-crowd.org/

DONE:

• basically functionality
• compute sentiment on n-grams first, if any found
• Make sentiment analysis exposed to AL interface

TODO:

• weight sentiment features by inverse frequency of words/terms in news agenda!? ('вылечились от короновируса' = 'cured from coronavirus')
• custom user-specific lexicons (hierarchy of custom "subgraphs" with lexicons in hierarchy of extensions)
• context-specific lexicons based on broad topic and/or sentence contexts
• context-specific sentiment based on topic location and its surroundings within a sentence
• More: https://blog.singularitynet.io/aigents-sentiment-detection-personal-and-social-relevant-news-be989d73b381

P.S.: File merging tips: https://stackoverflow.com/questions/4366533/how-to-remove-the-lines-which-appear-on-file-b-from-another-file-a

[akolonin]

In order to get sentiment computed on backend, need to add four files to installation root: https://github.com/aigents/aigents-java/commit/574e6a16050b87a7535ce45b0c4df1611f0f5d87 To get per-news-item sentiment updated, need:

1. Call "you think!" to enforce sentiment update in a news feed
2. Retrieve sentiment with request like "What new true sources, text, times, trust, relevance, social relevance, image, is, sentiment?"