NewsWeeder: Learning to Filter Netnews

By Ken Lang. 1995

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Natural Language Processing

Citations in the corpus

Semi-supervised Sequence Learning (2015, methodology): “In our experiments with document classification tasks with 20 Newsgroups [17] and DBpedia [20], and sentiment analysis with IMDB [22] and Rotten Tomatoes [26], LSTMs pretrained by recurrent language models or sequence autoencoders are usually better than LSTMs initialized randomly.”
Pretrained Transformers Improve Out-of-Distribution Robustness (2020, methodology): “For our OOD examples, we use validation examples from 20 Newsgroups (20 NG) (Lang, 1995), the English source side of English-German WMT16 and English-German Multi30K (Elliott et al.”
Measuring Association Between Labels and Free-Text Rationales (2020, background): “Turning to exceptions that focus on NL rationales; Latcinnik and Berant (2020) train a differentiable I→R;IR→O pipeline on CommonsenseQA, controlling for the complexity of the IR→O model to increase the likelihood that the model is faithful to the rationale, and ablating input I at inference to test”
A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks (2016, background): “In the 20 Newsgroups dataset (Lang, 1995), there are 20 different newsgroup subjects with a total of 20000 documents for the whole dataset.”
Collaborative Deep Learning for Recommender Systems (2014, methodology): “Content-based methods [17] make use of user profiles or product descriptions for recommendation.”
Joint Topic-Semantic-aware Social Recommendation for Online Voting (2017, methodology): “Content-based methods [14, 34] make use of user profiles or item descriptions as features for recommendation.”
Machine Learning Models that Remember Too Much (2017, background): “20 Newsgroups is a corpus of 20,000 documents classified into 20 categories [44].”
The Bayesian Case Model: A Generative Approach for Case-Based Reasoning and Prototype Classification (2014, methodology): “Figure 2: Prediction test accuracy reported for the Handwritten Digit [19] and 20 Newsgroups datasets [22].”
Understanding How Dimension Reduction Tools Work: An Empirical Approach to Deciphering t-SNE, UMAP, TriMAP, and PaCMAP for Data Visualization (2020, background): “Similarly to other studies of DR methods (Tang et al., 2016; McInnes et al., 2018; Amid and Warmuth, 2019; Belkina et al., 2019), we use the following datasets to evaluate and compare DR methods: COIL-20 (Nene et al., 1996b), COIL-100 (Nene et al., 1996a), S-curve with hole dataset (synthesized by t”
Achieving Forgetting Prevention and Knowledge Transfer in Continual Learning (2021, background): “This dataset [28] has 20 classes and each class has about 1000 documents.”
Regularization Paths for Generalized Linear Models via Coordinate Descent (2010, background): “The InternetAd and NewsGroup datasets are both sparse: 1% nonzero values for the former, 0.05% for the latter.”
Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift (2019, methodology): “• 20newsgroups (Lang, 1995) with even-numbered classes (10 classes out of 20) as in-distribution, oddnumbered classes as skewed data, and the One Billion Word Benchmark (LM1B) (Chelba et al.”
FLEX: Unifying Evaluation for Few-Shot NLP (2021, background): “• Class Transfer: FewRel [28], HuffPost [46], Amazon [30], 20News [38], and Reuters [41] take part in meta-training and meta-testing but with different classes.”
Communication-Efficient Distributed Optimization of Self-Concordant Empirical Loss (2015, background): “These datasets are selected to cover different relations between the sample size N = mn and the feature dimensionality d: N ≫ d (Covtype [6]), N ≈ d (RCV1 [25]) and N ≪ d (News20 [22, 23]).”