A biterm topic model for short texts

Yan, Xiaohui; Guo, Jiafeng; Lan, Yanyan; Cheng, Xueqi

doi:10.1145/2488388.2488514

Cited by 943 publications

(725 citation statements)

References 26 publications

Supporting

Mentioning

659

Contrasting

Unclassified

Order By: Relevance

“…We perform two sets of experiments to compare accuracy and scalability of our algorithm with two approaches namely LDA [13] 1 as the baseline and BTM [9] 2 as the current best known proposal to TD. For accuracy, we use the SNAP-Twitter 3 dataset containing trending topics from 2009 and evaluate the results using B-Cubed [14] score.…”

Section: Contributionmentioning

confidence: 99%

“…Yan et al proposed a different solution, called BiTerm (BTM) [9], where they tried to tackle the sparsity using an innovative bi-term document modeling approach. Authors modeled documents as a mixture of multiple bi-terms rather than single words.…”

Section: Related Workmentioning

confidence: 99%

“…We developed two sets of experiments to compare the efficiency and scalability of DeGPar against two state-ofthe-art approaches, LDA [5] and BTM [9]. We run each experiment 100 times on each algorithm and report the average result.…”

Section: A Experimental Settingmentioning

confidence: 99%

“…Afterwards, the algorithm proceeds by extracting those patterns by partitioning the graph using a novel node-cut partitioning algorithm. Through a set of experiments, we examined the accuracy and scalability of our algorithm against two state-of-the-art approaches LDA [5] and BTM [9]. The results show that our algorithm outperforms both approaches by an order of magnitude while exhibiting similar accuracy.…”

Section: B Scalabilitymentioning

confidence: 99%

“…For example [7] and [8] proposed to aggregate short texts either with higher similarity or from the same publisher but their experiments do not exceed 2 million documents. Another solution, called BiTerm (BTM) [9], proposed to assign topic probabilities over bi-grams rather than single words and learn topics using aggregated bi-gram patterns over the entire dataset rather than single documents. The solution, although shown successful to overcome sparsity on over 4 million documents, still remained comparable with LDA on it's execution time.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

DeGPar: Large Scale Topic Detection Using Node-Cut Partitioning on Dense Weighted Graphs

Ghoorchian

Girdzijauskas

Rahimian

2017

2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)

View full text Add to dashboard Cite

Abstract-Topic Detection (TD) refers to automatic techniques for locating topically related material in web documents [1]. Nowadays, massive amounts of documents are generated by users of Online Social Networks (OSNs), in form of very short text, tweets and snippets of news. While topic detection, in its traditional form, is applied to a few documents containing a lot of information, the problem has now changed to dealing with massive number of documents with very little information. The traditional solutions, thus, fall short either in scalability (due to huge number of input items) or sparsity (due to insufficient information per input item). In this paper we address the scalability problem by introducing an efficient and scalable graph based algorithm for TD on short texts, leveraging dimensionality reduction and clustering techniques. We first, compress the input set of documents into a dense graph, such that frequent cooccurrence patterns in the documents create multiple dense topological areas in the graph. Then, we partition the graph into multiple dense sub-graphs, each representing a topic. We compare the accuracy and scalability of our solution with two state-of-the-art solutions (including the standard LDA, and BiTerm). The results on two widely used benchmark datasets show that our algorithm not only maintains a similar or better accuracy, but also performs by an order of magnitude faster than the state-of-the-art approaches.

show abstract

Section: Contributionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: A Experimental Settingmentioning

confidence: 99%

Section: B Scalabilitymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

DeGPar: Large Scale Topic Detection Using Node-Cut Partitioning on Dense Weighted Graphs

Ghoorchian

Girdzijauskas

Rahimian

2017

2017 IEEE 37th International Conference on Distributed Computing Systems (ICDCS)

View full text Add to dashboard Cite

show abstract

Document Clustering

Anastasiu¹,

Tagarelli²

2017

Wiley StatsRef: Statistics Reference Online

View full text Add to dashboard Cite

In a world flooded with information, document clustering is an important tool that can help categorize and extract insight from text collections. It works by grouping similar documents, while simultaneously discriminating between groups. In this article, we provide a brief overview of the principal techniques used to cluster documents and introduce a series of novel deep‐learning‐based methods recently designed for the document clustering task. In our overview, we point the reader to salient works that can provide a deeper understanding of the topics discussed.

show abstract

Identifying social roles using heterogeneous features in online social networks

Liu

Sun

et al. 2019

Asso for Info Science & Tech

View full text Add to dashboard Cite

Role analysis plays an important role when exploring social media and knowledge-sharing platforms for designing marking strategies. However, current methods in role analysis have overlooked content generated by users (e.g., posts) in social media and hence focus more on user behavior analysis. The user-generated content is very important for characterizing users. In this paper, we propose a novel method which integrates both user behavior and posted content by users to identify roles in online social networks. The proposed method models a role as a joint distribution of Gaussian distribution and multinomial distribution, which represent user behavioral feature and content feature respectively. The proposed method can be used to determine the number of roles concerned automatically. The experimental results show that the proposed method can be used to identify various roles more effectively and to get more insights on such characteristics.

show abstract

A biterm topic model for short texts

Cited by 943 publications

References 26 publications

DeGPar: Large Scale Topic Detection Using Node-Cut Partitioning on Dense Weighted Graphs

DeGPar: Large Scale Topic Detection Using Node-Cut Partitioning on Dense Weighted Graphs

Document Clustering

Identifying social roles using heterogeneous features in online social networks

Contact Info

Product

Resources

About