Topic Detection and Tracking Pilot Study Final Report

Allan, James; Carbonell, Jaime G.; Doddington, George R.; Yamron, Jonathan; Yang, Yiming

doi:10.1184/r1/6626252.v1

Cited by 243 publications

(4 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Automatically identifying topical changes within the document set requires methods to extract machine-readable topics from the collection. Topic modeling provides a statistical approach to discover topics within a given corpus, 1 https://www.microsoft.com/en-us/research/project/microsoft-academic-graph/…”

Section: Identifying the Evolution Of Topicsmentioning

confidence: 99%

“…First story detection (FSD) is one of the parts of TDT research tasks. The goal of FSD is to search and organize new topics from multilingual news articles or identify the first article introducing the new story [1]. Topic-conditioned FSD with a supervised learning algorithm first classified news articles into a set of pre-defined topic categories before identifying novelty within each topic [38].…”

Section: Identifying and Predicting New Topicsmentioning

confidence: 99%

See 1 more Smart Citation

Analyzing the generalizability of the network-based topic emergence identification method

Jung

Segev

2022

View full text Add to dashboard Cite

Topic evolution helps the understanding of current research topics and their histories by automatically modeling and detecting the set of shared research fields in academic publications as topics. This paper provides a generalized analysis of the topic evolution method for predicting the emergence of new topics, which can operate on any dataset where the topics are defined as the relationships of their neighborhoods in the past by extrapolating to the future topics. Twenty sample topic networks were built with various fields-of-study keywords as seeds, covering domains such as business, materials, diseases, and computer science from the Microsoft Academic Graph dataset. The binary classifier was trained for each topic network using 15 structural features of emerging and existing topics and consistently resulted in accuracy and F1 over 0.91 for all twenty datasets over the periods of 2000 to 2019. Feature selection showed that the models retained most of the performance with only one-third of the tested features. Incremental learning was tested within the same topic over time and between different topics, which resulted in slight performance improvements in both cases. This indicates there is an underlying pattern to the neighbors of new topics common to research domains, likely beyond the sample topics used in the experiment. The result showed that network-based new topic prediction can be applied to various research domains with different research patterns.

show abstract

Section: Identifying the Evolution Of Topicsmentioning

confidence: 99%

Section: Identifying and Predicting New Topicsmentioning

confidence: 99%

Analyzing the generalizability of the network-based topic emergence identification method

Jung

Segev

2022

View full text Add to dashboard Cite

show abstract

“…Primarily, any such system has four major processing phases: gathering raw data, preprocessing, term weight estimation, and grouping documents based on their similarity. Traditionally, events were detected by gathering [3] historical data such as news articles, radio broadcast articles, and classified columns. Due to the advent of social media streams, an enormous amount of real-time data has become obtainable [1].…”

Section: Background and Related Workmentioning

confidence: 99%

“…Discovering topics from a document, or any kind of contextual source of information was initiated by topic detection and tracking (TDT) [3] project. This concept further moved on to integrate the time-ISSN: 2252-8776  varying property [12] by incorporating online data from social streams.…”

Section: Online Event Detectionmentioning

confidence: 99%

Meliorating usable document density for online event detection

Samanta

Meena²,

Mazumdar³

et al. 2022

IJ-ICT

View full text Add to dashboard Cite

<span>Online event detection (OED) has seen a rise in the research community as it can provide quick identification of possible events happening at times in the world. Through these systems, potential events can be indicated well before they are reported by the news media, by grouping similar documents shared over social media by users. Most OED systems use textual similarities for this purpose. Similar documents, that may indicate a potential event, are further strengthened by the replies made by other users, thereby improving the potentiality of the group. However, these documents are at times unusable as independent documents, as they may replace previously appeared noun phrases with pronouns, leading OED systems to fail while grouping these replies to their suitable clusters. In this paper, a pronoun resolution system that tries to replace pronouns with relevant nouns over social media data is proposed. Results show significant improvement in performance using the proposed system.</span>

show abstract