JohanyCheung#notebook#冷启动推荐1

## 冷启动 冷启动就需要`user`或者`item`的特征来配合了, 仅使用协同过滤的FM模型是无法做到冷启动的. 我们选择[StackExchange dataset](https://archive.org/details/stackexchange)作为数据集, 这个数据集除了行为交互数据, 还有**item feature**, 因此可以做关于`item`的冷启动. ```python import numpy as np from lightfm.datasets import fetch_stackexchange data = fetch_stackexchange('crossvalidated', test_set_fraction=0.1, indicator_features=False, tag_features=True) data ``` ``` {'train': <3221x72360 sparse matrix of type '' with 57830 stored elements in COOrdinate format>, 'test': <3221x72360 sparse matrix of type '' with 4307 stored elements in COOrdinate format>, 'item_features': <72360x1246 sparse matrix of type '' with 198963 stored elements in Compressed Sparse Row format>, 'item_feature_labels': array(['bayesian', 'prior', 'elicitation', ..., 'events', 'mutlivariate', 'sample-variance'], dtype='