Metadata-Version: 2.1
Name: lexicon-overlap-score
Version: 0.0.1
Summary: Functions for calculation of the lexicon overlap score
Home-page: https://gitlab.rrz.uni-hamburg.de/bay1620/lexicon_overlap_score
Author: Felix Welter
Author-email: felixwelter@gmail.com
License: UNKNOWN
Description: # Lexicon overlap score
        The lexicon overlap score is a similarity measure for sentiment lexicons as they are used in Sentiment Classification. 
        
        # Installation
        Install via `pip install lexicon_overlap_score`
        
        # Usage
        Import via `import lexicon_overlap_score as los`
        
        Your lexicons should be a pandas dataframe with the columns `word` and `so` (semantic orientation).
        Then the following three functions can be used.
        
        ``` python
        los.simple(lex1, lex2)
        los.binary(lex1, lex2)
        los.score(lex1, lex2)
        ```
        
        `simple` and `binary` return a value between 0 and 1, while `score` returns values between -1 and 1.
        The functions expect positive words to have a positive value and negative words to have a negative value.
        
        If your lexicon is a python dictionary mapping words to values, convert it to a pandas dataframe. 
        
        ``` python
        import pandas as pd
        df_lex = pd.DataFrame().from_dict(dict_lex, orient="index").reset_index().rename(columns={"index": "word", 0: "so"})
        ```
        
        # Example
        ```
        import pandas as pd
        import lexicon_overlap_score as los
        df1 = pd.DataFrame([("test",   1), ("free",    1), ("other", 1), ("check", 0.5)], columns=["word", "so"])
        df2 = pd.DataFrame([("test", 0.5), ("free",    1)],                               columns=["word", "so"])
        los.simple(df1, df2), los.binary(df1, df2), los.score(df1, df2)
        ```
        
Platform: UNKNOWN
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.6
Description-Content-Type: text/markdown
