sub:assertion {
this: dcterms:description """---
title: \"[[EVD]] - increased word ambiguity in abstracts was associated with slightly lower modularity of citation networks for those abstracts - [[@mcmahanAmbiguityEngagement2018]]\"
url: https://roamresearch.com/#/app/megacoglab/page/0dCO_io-p
author: Joel Chan
date: Mon Jul 12 2021 12:35:18 GMT-0400 (Eastern Daylight Time)
---
- Summary::
- Increased word [ambiguity]([[ambiguity]]
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FYK5hjRvILZ.png?alt=media&token=91149ef7-7b6d-4595-ba46-553a1e300607) (p. 891)
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2Ff1KNxVhp1L.png?alt=media&token=360f4488-5542-48a5-b832-624d4cecda87) (p. 892, Table 2)
- Note: this effect was ~order of magnitude smaller than effects of number of disciplines and subject-wide modularity
- **Grounding Context**
- sample: ~1.1M abstracts from Web of Science (after trimming for min. threshold citations, and at least 5 usable terms for ambiguity measure in abstract)
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2F6swfV2e5lD.png?alt=media&token=663be791-00fa-43a8-939e-16b1e2b3839b) (p. 880 - 881)
- [[ambiguity]]
- define as posterior distribution over pr(M | t, c) (probability of each possible meaning for a given token in a given context) - see Appendix A (how to build \"meaning graph\" for each token), and Appendix B (how to estimate ambiguity)
- very similar to [[LDA]]
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FkBZZ_HTTGv.png?alt=media&token=1b9f4a7c-5f25-4555-82c0-4f2785d46fb6) (p. 898)
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FyJ2BvBP1aJ.png?alt=media&token=06e19e17-993e-43dc-9095-47215d99ff63)
- meaning simplified to proxy into synonyms from Moby and WordNet thesauri
- would prefer to define as cliques in synonym graph, but... for reasons, decided to simplify to just directed edges + the neighborhood
- examples (from data)
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FjNbxETh3nc.png?alt=media&token=7bf4a619-0ca2-4960-81e6-cc1096a24ba9)
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FgI8AhKGeTl.png?alt=media&token=ae39c2be-5b06-48f4-bb34-4216808fc838) (p. 885)
- validated against human judgments (see Appendix C)
- \"engagement\" operationalized as graph entropy over subgraph from each article for max (1k cites, 3 recursive runs), computed in terms of maximal [[network modularity]]
- toy example:
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FAUmk1jo2Ua.png?alt=media&token=c9ed7651-5fea-44fe-994a-c769e70cb68d)
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FzWrx4znKTB.png?alt=media&token=ac0e15f3-3cfc-4454-82ab-bf0acb1a3689) (p. 880)
- examples (from data):
- ![](https://firebasestorage.googleapis.com/v0/b/firescript-577a2.appspot.com/o/imgs%2Fapp%2Fmegacoglab%2FKUb7aKx_yi.png?alt=media&token=d9e03062-c18f-432c-896a-32a3062151ed)
- network modularity correlated at -.76 with topological entropy for random sample of 1.2k abstracts
- ---
- [[ACL Recontextualizing Claims and Evidence Shared Task]]
- Status:: #resultGrounded
- Annotator::
- ResultGrounding:: [[table]]
""" ;
a "https://w3id.org/kpxl/gen/terms/Evidence" ;
rdfs:label "[[EVD]] - increased word ambiguity in abstracts was associated with slightly lower modularity of citation networks for those abstracts - [[@mcmahanAmbiguityEngagement2018]]" ;
foaf:page "https://roamresearch.com/megacoglab/page/0dCO_io-p" .
}