添加了代码文件

Project1/related.py

+import heapq
+import numpy as np
+
+# 利用priority queue性质求解Top-K问题
+def topk(inputs, k):
+    pq = []
+    pq_index = []
+    for index, value in enumerate(inputs):
+        if len(pq) < k:
+            heapq.heappush(pq, value)
+            heapq.heappush(pq_index, index)
+        elif value > pq[0]:
+            heapq.heapreplace(pq, value)
+            heapq.heapreplace(pq_index, index)
+    ret = list()
+    while pq_index:
+        ret.append(heapq.heappop(pq_index))
+  
+    return ret[::-1]
+
+
+# 计算每个词的glove词向量的语义相似度，每个词取最相似topk
+def get_top_glove_results(vec, emb, vocab):
+
+    # 计算余弦相似度
+    cos = np.dot(vec, emb) / (np.linalg.norm(vec)*(np.linalg.norm(emb)))
+    inputs = list(cos)
+    k = 10
+    top_idxs = topk(inputs, k)  # top_idxs存放相似度最高的词向量下标
+                                
+    return vocab[top_idxs]  # 返回相似度最高的问题对应的答案，作为TOP10答案
+
+
+
+with open('./data/word2id.txt', 'r') as f:
+    word2id = eval(f.read())
+
+emb = np.loadtxt('./data/embedding.txt',delimiter=',')
+
+res = {}
+vocab = np.array(list(word2id.keys()))
+for k in word2id:
+    word_vec = emb[word2id[k]-1, :]
+    res[k] = list(get_top_glove_results(word_vec, emb.T, vocab))
+
+# 保存结果
+with open('related_words.txt', 'w') as f:
+    f.write(str(res))
+    
+        
\ No newline at end of file