RankFlow

Qin, Jiarui; Zhu, Jiachen; Chen, Bin; Liu, Zhongfan; Liu, Weiwen; Tang, Ruiming; Zhang, Rui; Yu, Yong; Zhang, Weinan

doi:10.1145/3477495.3532050

Cited by 14 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Pair-wise contrastive learning. Most of the existing retrieval methods (Qin et al, 2022;Hossain et al, 2020) obtain the relevant documents by ranking the candidate documents based on the similarity to the query. They only minimize the negative log-likelihood of the positive documents, implicitly optimizing their model to binary classify the documents.…”

Section: Multi-granularity Contrastive Learningmentioning

confidence: 99%

Towards a Unified Framework for Reference Retrieval and Related Work Generation

Shi,

Gao,

Zhang

et al. 2023

Findings of the Association for Computational Linguistics: EMNLP 2023

View full text Add to dashboard Cite

The task of related work generation aims to generate a comprehensive survey of related research topics automatically, saving time and effort for authors. Existing methods simplify this task by using human-annotated references in a large-scale scientific corpus as information sources, which is time-and cost-intensive. To this end, we propose a Unified Reference Retrieval and Related Work Generation Model (UR 3 WG), which combines reference retrieval and related work generation processes in a unified framework based on the large language model (LLM). Specifically, UR 3 WG first leverages the world knowledge of LLM to extend the abstract and generate the query for the subsequent retrieval stage. Then a lexicon-enhanced dense retrieval is proposed to search relevant references, where an importance-aware representation of the lexicon is introduced. We also propose multi-granularity contrastive learning to optimize our retriever. Since this task is not simply summarizing the main points in references, it should analyze the complex relationships and present them logically. We propose an instruction-tuning method to guide LLM to generate related work. Extensive experiments on two wide-applied datasets demonstrate that our UR 3 WG outperforms the state-of-the-art baselines in both generation and retrieval metrics.

show abstract

Section: Multi-granularity Contrastive Learningmentioning

confidence: 99%

Towards a Unified Framework for Reference Retrieval and Related Work Generation

Shi,

Gao,

Zhang

et al. 2023

Findings of the Association for Computational Linguistics: EMNLP 2023

View full text Add to dashboard Cite

show abstract

“…Following Rieder et al’s (2018) mixed-methods methodology to study patterns of change over time on social media, we generated files that contained the top 20 most shared unique YouTube videos on Twitter every day over the period of 4 months (February–May 2020). To visualise our data and identify patterns in how users shared YouTube videos on Twitter, we used the RankFlow tool (Rieder, 2016) – which creates flow diagrams that make changes over time easily observable – and created rank flow morphologies for every month of the study period. Since we wanted to identify patterns of ‘inauthentic’ behaviour in how YouTube content creators share videos on Twitter (our RQ2-A), we were interested not only in how many times a YouTube video was shared on any given day, but in the relationship between the number of times a video was tweeted and the number of unique users that shared that video.…”

Section: Introductionmentioning

confidence: 99%

Acting like a bot as a defiance of platform power: Examining YouTubers’ patterns of ‘inauthentic’ behaviour on Twitter during COVID-19

Matamoros-Fernández,

Bartolo,

Alpert

2024

New Media & Society

View full text Add to dashboard Cite

This article examines YouTubers’ ‘bot like’ behaviour on Twitter and conceptualises it as a defiance of platform power in delimiting the boundaries of ‘authenticity’. This entrepreneurial capture of ‘botness’ is understudied and deserves attention. We focus on a platform with a clear monetisation scheme, YouTube, and on patterns of ‘inauthentic’ behaviour in how people shared YouTube videos on Twitter during the early stages of the COVID-19 pandemic (February–May 2020). The global coronavirus crisis forced social media platforms to take unprecedented (mostly automated) steps to moderate content and to introduce new policies on ‘appropriate’ conduct, which we argue may have an impact on emerging social media content creators’ attempts to boost visibility online.

show abstract