Weakly-supervised acquisition of labeled class instances using graph random walks

Talukdar, Partha; Reisinger, Joseph; Paşca, Marius; Ravichandran, Deepak; Bhagat, Rahul; Pereira, Fernando

doi:10.3115/1613715.1613787

Cited by 80 publications

(65 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We are less concerned with extraction performance, but focus on the accuracy of the learned type system by measuring how well it performs in a prediction task. Talukdar et al (2008) and Talukdar and Pereira (2010) present graph-based approaches to the similar problem of class-instance learning. While this provides a way to discover types, it requires a large graph that does not easily generalize to new instances (transductive), since it produces no predictive model.…”

Section: Related Workmentioning

confidence: 99%

How Well can We Learn Interpretable Entity Types from Text?

Hovy¹

2014

Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)

View full text Add to dashboard Cite

Many NLP applications rely on type systems to represent higher-level classes. Domain-specific ones are more informative, but have to be manually tailored to each task and domain, making them inflexible and expensive. We investigate a largely unsupervised approach to learning interpretable, domain-specific entity types from unlabeled text. It assumes that any common noun in a domain can function as potential entity type, and uses those nouns as hidden variables in a HMM. To constrain training, it extracts co-occurrence dictionaries of entities and common nouns from the data. We evaluate the learned types by measuring their prediction accuracy for verb arguments in several domains. The results suggest that it is possible to learn domain-specific entity types from unlabeled data. We show significant improvements over an informed baseline, reducing the error rate by 56%.

show abstract

Section: Related Workmentioning

confidence: 99%

How Well can We Learn Interpretable Entity Types from Text?

Hovy¹

2014

Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)

View full text Add to dashboard Cite

show abstract

“…Entity clustering from semi-structured data has been addressed previously [18,23,9]. These approaches however do not address the issue of mixed-membership.…”

Section: Related Workmentioning

confidence: 99%

From Topic Models to Semi-supervised Learning: Biasing Mixed-Membership Models to Exploit Topic-Indicative Features in Entity Clustering

Balasubramanyan

Dalvi

Cohen

2013

Advanced Information Systems Engineering

View full text Add to dashboard Cite

Abstract. We present methods to introduce different forms of supervision into mixed-membership latent variable models. Firstly, we introduce a technique to bias the models to exploit topic-indicative features, i.e. features which are apriori known to be good indicators of the latent topics that generated them. Next, we present methods to modify the Gibbs sampler used for approximate inference in such models to permit injection of stronger forms of supervision in the form of labels for features and documents, along with a description of the corresponding change in the underlying generative process. This ability allows us to span the range from unsupervised topic models to semi-supervised learning in the same mixed membership model. Experimental results from an entity-clustering task demonstrate that the biasing technique and the introduction of feature and document labels provide a significant increase in clustering performance over baseline mixed-membership methods.

show abstract

“…Now, analyzing the objective in Equation 12 in the manner outlined in Section 4, we arrive at the update rule shown in Equation 13.…”

Section: Extensions: Non-mutually Exclusive Labelsmentioning

confidence: 99%

“…Most algorithms only output label information to the unlabeled data in a transductive setting, while some algorithms are designed for the semi-supervised framework and build a classification model which can be applied to out-of-sample examples. Adsorption [1] is one such recently proposed graph based semi-supervised algorithm which has been successfully used for different tasks, such as recommending YouTube videos to users [1] and large scale assignment of semantic classes to entities within Information Extraction [13]. Adsorption has many desirable properties: it can perform multiclass classification, it can be parallelized and hence can be scaled to handle large data sets which is of particular importance for semi-supervised algorithms.…”

Section: Introductionmentioning

confidence: 99%

New Regularized Algorithms for Transductive Learning

Talukdar

Crammer

2009

Lecture Notes in Computer Science

Self Cite

130

141

View full text Add to dashboard Cite

Abstract. We propose a new graph-based label propagation algorithm for transductive learning. Each example is associated with a vertex in an undirected graph and a weighted edge between two vertices represents similarity between the two corresponding example. We build on Adsorption, a recently proposed algorithm and analyze its properties. We then state our learning algorithm as a convex optimization problem over multi-label assignments and derive an efficient algorithm to solve this problem. We state the conditions under which our algorithm is guaranteed to converge. We provide experimental evidence on various real-world datasets demonstrating the effectiveness of our algorithm over other algorithms for such problems. We also show that our algorithm can be extended to incorporate additional prior information, and demonstrate it with classifying data where the labels are not mutually exclusive.

show abstract

Weakly-supervised acquisition of labeled class instances using graph random walks

Cited by 80 publications

References 10 publications

How Well can We Learn Interpretable Entity Types from Text?

How Well can We Learn Interpretable Entity Types from Text?

From Topic Models to Semi-supervised Learning: Biasing Mixed-Membership Models to Exploit Topic-Indicative Features in Entity Clustering

New Regularized Algorithms for Transductive Learning

Contact Info

Product

Resources

About