Quantifying the Effects of Text Duplication on Semantic Models

Schofield, Alexandra; Thompson, Laure; Mimno, David

doi:10.18653/v1/d17-1290

Cited by 24 publications

(18 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This characteristic is largely absent in data collected for the Internet Archive, where only 22 tweets were authored by @internetarchive, of which nine were original tweets. The sensitivity of latent dirichlet allocation (LDA) topic modeling to repeated text has been turned to advantage in the topic model of the aggregated corpus to reveal the most retweeted tweets, many of which originated from the archives’ accounts (Schofield, Thompson, & Mimno, ). Repeated text in this study has been interpreted as a measure of user interest in the subject matter.…”

Section: Methodsmentioning

confidence: 99%

A twitter case study for assessing digital sound

Giannetti

2018

Asso for Info Science & Tech

View full text Add to dashboard Cite

Academic and cultural heritage institutions around the world have made measurable strides in the development of digital sound archives oriented towards research and access, but their impact on scholarship and society has been little studied. Traditionally, impact has been measured by citations; yet these are problematic metrics for nontraditional outputs like sound recordings. Social media data provide a promising avenue of investigation for measuring scholarly as well as societal impact. Twitter in particular has been shown to provide a high number of references for cultural and research outputs in all disciplines. This study analyzes Twitter references pertaining to the collections of five digital sound archives: British Library Sounds, Europeana Sounds, the Internet Archive Audio Archive, PennSound, and UbuWeb. By using text analysis methods to identify high‐frequency events and trends, and labeling them with a rubric designed for measuring the impact of digital heritage resources, this study provides preliminary insights on user values as they relate to digital sound collections. Despite the limitations of using social media data, the evidence gathered in this case study characterizes aspects of the use of digital sound collections, and may point to future priorities for the digital preservation of sound.

show abstract

Section: Methodsmentioning

confidence: 99%

A twitter case study for assessing digital sound

Giannetti

2018

Asso for Info Science & Tech

View full text Add to dashboard Cite

show abstract

“…To examine this phenomenon in isolation, we repeat the training corpus twice and observe the effect of diversity-aware strategies. The corpus duplication technique has been previously used to probe semantic models (Schofield et al, 2017). Figure 3 shows learning curves for strategies under the original and corpus duplication settings.…”

Section: Corpus Duplication Settingmentioning

confidence: 99%

Diversity-Aware Batch Active Learning for Dependency Parsing

Shi¹,

Benton²,

Malioutov³

et al. 2021

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

View full text Add to dashboard Cite

While the predictive performance of modern statistical dependency parsers relies heavily on the availability of expensive expert-annotated treebank data, not all annotations contribute equally to the training of the parsers. In this paper, we attempt to reduce the number of labeled examples needed to train a strong dependency parser using batch active learning (AL). In particular, we investigate whether enforcing diversity in the sampled batches, using determinantal point processes (DPPs), can improve over their diversity-agnostic counterparts. Simulation experiments on an English newswire corpus show that selecting diverse batches with DPPs is superior to strong selection strategies that do not enforce batch diversity, especially during the initial stages of the learning process. Additionally, our diversityaware strategy is robust under a corpus duplication setting, where diversity-agnostic sampling strategies exhibit significant degradation.

show abstract

“…Schofield et al [17] show learning on how repeated texts affect semantic models. They trained a Latent Dirichlet Allocation model and Latent Semantic Analysis model over different levels of repeated texts.…”

Section: Related Workmentioning

confidence: 99%

Detection of Mergeable Wikipedia Articles Utilizing Multiple Similarity Measures

Wang

Iwaihara

2020

Journal of Information Processing

View full text Add to dashboard Cite

Wikipedia is the largest online encyclopedia, in which articles are edited by different volunteers with different thoughts and styles. Sometimes two or more articles' titles are different but the themes of these articles are exactly the same or strongly similar. Administrators and editors are supposed to detect such article pairs and determine whether they should be merged together. We call an article pair is mergeable if it is discussed for possible merge, and a merged article pair is such that the pair is actually merged. In this paper, we propose a method to automatically determine whether an article pair is mergeable or merged. According to Wikipedia Guidelines for article merge, in the duplicate case, the article pairs are covering exactly the same contents. In the overlap case, the article pairs are covering related subjects that have a significant overlap. The content of an overlapped part is similar but the words in the pair can be extensively different, so methods that exploit semantic relatedness are necessary. We consider various textual similarities and semantic relatedness. For integrating word embeddings on the target dataset and the global large corpus, we propose linear and non-linear combinations of multiple embedding results and rebuilding word vectors for evaluating semantic relatedness. We clarify the differences between our method and previous researches for combining multiple word embeddings. We also deal with overlap cases by computing Jaccard similarity between article pairs. We combine Jaccard similarity, common-link article count and word embedding-based relatedness together, to predict whether the article pair should be merged. We explore the relationship between segment-level (paragraph-level) similarity and mergeable/merged article pairs, then propose Multimodal Similarity-Based Merge Prediction (MSBMP) which combines the proposed new features by Random Forest, to predict mergeable/merged article pairs. Our evaluations are performed on real mergeable and merged article pairs. Remarkable superiorities of MSBMP are shown, with apparent improvement from baselines of WikiSearch, TFIDF and word embeddings.

show abstract

Quantifying the Effects of Text Duplication on Semantic Models

Cited by 24 publications

References 19 publications

A twitter case study for assessing digital sound

A twitter case study for assessing digital sound

Diversity-Aware Batch Active Learning for Dependency Parsing

Detection of Mergeable Wikipedia Articles Utilizing Multiple Similarity Measures

Contact Info

Product

Resources

About