Bag of Tricks for Efficient Text Classification

This paper explores a simple and efficient baseline for text classification. Our experiments show that our fast text classifier fastText is often on par with deep learning classifiers in terms of accuracy, and many orders of magnitude faster for training and evaluation. We can train fastText on more than one billion words in less than ten minutes using a standard multicore~CPU, and classify half a million sentences among~312K classes in less than a minute.

PDF Abstract EACL 2017 PDF EACL 2017 Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Text Classification AG News fastText Error 7.5 # 13
Sentiment Analysis Amazon Review Full FastText Accuracy 60.2 # 8
Sentiment Analysis Amazon Review Polarity FastText Accuracy 94.6 # 8
Emotion Recognition in Conversation CPED FastText Accuracy of Sentiment 48.62 # 7
Macro-F1 of Sentiment 30.33 # 11
Text Classification DBpedia FastText Error 1.4 # 18
Sentiment Analysis Sogou News fastText, h=10, bigram Accuracy 96.8 # 1
Text Classification Yahoo! Answers FastText Accuracy 72.3 # 9
Sentiment Analysis Yelp Binary classification fastText, h=10, bigram Error 4.3 # 17
Sentiment Analysis Yelp Fine-grained classification FastText Error 36.1 # 14

Methods


No methods listed for this paper. Add relevant methods here