Prophet

prophet is a secret project now, I won't tell you what it is until the conpetition finish.

• it seems feasible that use the kmeans algorithm as the hash function of LSH.

• 获取dts特征

• dts映射为向量

• kmeans哈希

• 计算距离

• 输出结果

• TFIDF:

1. 获取词袋，统计idf；
2. 获取tfidf
3. 计算pca

• word2vec：

1. 计算word2vec，获取词袋
2. 计算文本向量（1，word2vec相加求平均）