cbow_logic.py

# cbow_logic.py
import gensim
import os
import argparse
from typing import List, Tuple
import shlex


class MeaningCalculator:
    def __init__(self, model_path: str = "/models/cbow/cbow_model.kv"):
        if not os.path.exists(model_path):
            raise FileNotFoundError(f"Model not found at: {model_path}")
        self.model = gensim.models.KeyedVectors.load(model_path, mmap='r')

    def evaluate_expression(self, expression: str, topn: int = 10) -> List[Tuple[str, float]]:
    # Evaluate expressions like '"new york" - city + capital'.
        tokens = shlex.split(expression)  # Handles quoted terms properly
        positive = []
        negative = []
        current_op = "+"

        for token in tokens:
            print(token)
            if token in ["+", "-"]:
                current_op = token
            else:
                if current_op == "+":
                    positive.append(token)
                else:
                    negative.append(token)

        try:
            return self.model.most_similar(positive=positive, negative=negative, topn=topn)
        except KeyError as e:
            return [("InputError", 0.0)] 

from gensim.models import KeyedVectors



if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Evaluate word vector expressions using CBOW.")
    parser.add_argument("expression", type=str, help="Expression like 'king - man + woman'")
    parser.add_argument("--model_path", type=str, default="./models/cbow_model.kv", help="Path to CBOW model")
    args = parser.parse_args()

    calc = MeaningCalculator(model_path=args.model_path)
    results = calc.evaluate_expression(args.expression)

    print(f"\nExpression: {args.expression}\nTop Results:")
    for word, score in results:
        print(f"  {word:<15} {score:.4f}")