jaccard_sim_ngram (string/string_sim.py)

Provides Jaccard n-gram similarity and Levenshtein distance between strings.

# Source: https://github.com/crazyyanchao/llmcompiler | License: Apache-2.0 | Copyright You must retain, in the Source form of any Derivative Works
# Provided via justclaw.it.com; keep this notice with the code.
# -*- coding: utf-8 -*-
"""
@Author  : Yc-Ma
@Desc    : LLMCompiler
@Time    : 2024-08-02 09:30:49
"""


def jaccard_sim_ngram(answer, ref, ngram=2):
    if answer in ref: return 1.0
    if len(answer) - ngram > 0:
        answer_ngrams = [answer[i: i + ngram] for i in range(0, len(answer) - ngram + 1)]
    else:
        answer_ngrams = answer  # 退化到unigram
    if len(ref) - ngram > 0:
        ref_ngrams = [ref[i: i + ngram] for i in range(0, len(ref) - ngram + 1)]
    else:
        ref_ngrams = ref  # 退化到unigram
    intersection = len(set(answer_ngrams).intersection(set(ref_ngrams

... (truncated -- full source via MCP)