Create a simple sentiment annotator for named entities

[roddar92]

• Chose dataset with NE: Person, Organization, ... (https://tatianashavrina.github.io/2018/08/30/datasets/#ner);
• Apply two of the our best models for NE's context (N = 5-7 words from the left + 5-7 words from the right);
• Each best model for each student :) One model should be a simple, another model - a neural;

Note: try to design an architecture prototype of our annotator!

Hints: main classes (Annotator(Model model), Model (abstract class)), Handler

Naive architecture:

• Application() (main point into our program, Annotator will be called here),
• Annotator(Model), main method annotate(Dataset) --> NERConcordance, т.е. на входе - размеченный датасет Dataset, на выходе - таблица с результатами NERConcordance),
• Model(наша сентимент-модель),
• Dataset(размеченный датасет с BIO-аннотацией, список последовательностей Text),
• Text(текст с разметкой BIO для каждого токена, т.е. список размеченных токенов),
• NERConcordance(список SentimentNER, ocontext_window_size=5),
• SentimentNER(left context, entity, right context, sentiment_score),