| 271 | |
| 272 | |
| 273 | def analogy(pos1, neg1, pos2, neg2): |
| 274 | # don't actually use pos2 in calculation, just print what's expected |
| 275 | print("testing: %s - %s = %s - %s" % (pos1, neg1, pos2, neg2)) |
| 276 | for w in (pos1, neg1, pos2, neg2): |
| 277 | if w not in word2idx: |
| 278 | print("Sorry, %s not in word2idx" % w) |
| 279 | return |
| 280 | |
| 281 | p1 = W[word2idx[pos1]] |
| 282 | n1 = W[word2idx[neg1]] |
| 283 | p2 = W[word2idx[pos2]] |
| 284 | n2 = W[word2idx[neg2]] |
| 285 | |
| 286 | vec = p1 - n1 + n2 |
| 287 | |
| 288 | distances = pairwise_distances(vec.reshape(1, D), W, metric='cosine').reshape(V) |
| 289 | idx = distances.argsort()[:10] |
| 290 | |
| 291 | # pick the best that's not p1, n1, or n2 |
| 292 | best_idx = -1 |
| 293 | keep_out = [word2idx[w] for w in (pos1, neg1, neg2)] |
| 294 | for i in idx: |
| 295 | if i not in keep_out: |
| 296 | best_idx = i |
| 297 | break |
| 298 | |
| 299 | print("got: %s - %s = %s - %s" % (pos1, neg1, top_words[best_idx], neg2)) |
| 300 | print("closest 10:") |
| 301 | for i in idx: |
| 302 | print(top_words[i], distances[i]) |
| 303 | |
| 304 | print("dist to %s:" % pos2, cos_dist(p2, vec)) |
| 305 | |
| 306 | |
| 307 | analogy('king', 'man', 'queen', 'woman') |