Codenames clue giver

codenames.py

# coding: utf-8
from gensim.models import KeyedVectors
# Downloaded from fasttext: https://fasttext.cc/docs/en/english-vectors.html
# Converted to word2vec binary format for faster loading (see convert.py)
vec = KeyedVectors.load_word2vec_format("~/data/wiki-news-300d-1M.vec.bin", binary=True)
from itertools import combinations
from nltk.stem import WordNetLemmatizer,PorterStemmer

wnl = WordNetLemmatizer()
stemmer = PorterStemmer()

def isplural(word):
    lemma = wnl.lemmatize(word, 'n')
    plural = True if word is not lemma else False
    return plural

def isCap(word):
    return word != word.lower()

def isSame(a,b):
    return stemmer.stem(a) == stemmer.stem(b) or a != a.lower() or b != b.lower()


while True:
    # words is all the words on your team (red or blue words)
    words = input("words >> ").strip().split()
    # give one negative word (usually the black word)
    negword = input("neg >> ").strip()
    neg = [negword]
    
    print(words)
    print(neg)
    
    bestpair = None
    bestcand = ""
    bestscore = -1

    num = 2
    for p in combinations(words,num):
        print(p)
        sim = vec.most_similar(positive=p, negative=neg)
        for s in sim[:5]:
            cand = s[0]
            score = s[1]
            if isSame(cand, p[0]) or isSame(cand, p[1]):
                continue
            print(cand, score)
            if score > bestscore:
                bestpair = p
                bestscore = score
        print()

    print(bestpair)
    sim = vec.most_similar(positive=bestpair, negative=neg)
    for s in sim[:5]:
        cand = s[0]
        if isSame(bestpair[0], cand) or isSame(bestpair[1], cand):
            continue
        print(s)

convert.py

from gensim.models.keyedvectors import KeyedVectors

fname = "~/data/wiki-news-300d-1M.vec"
model = KeyedVectors.load_word2vec_format(fname, binary=False)
model.save_word2vec_format(fname + ".bin", binary=True)