Er‐Te Zheng

Zohreh Zahedi, Yanqing Zhang, Zekun Han et al.

Organisational accounts are an integral part of the Twitter (now X) ecosystem. This study identified 9,842 research- and policy-related organisational accounts that had tweeted about scholarly publications by linking three global organisational databases (GRID, ROR, and Overton) with two altmetric databases containing Twitter data (Altmetric and the former Crossref Event Data). The resulting openly available dataset was used to examine organisational activity in scholarly communication across three dimensions: social media capital, tweeting activity, and engagement level. The results show that, compared to all Twitter users engaged in scholarly communication, organisational accounts hold a notable advantage in terms of follower bases and the proportion of scholarly tweets. Their scholarly tweets achieve high visibility through likes and retweets but perform weakly in generating more conversational forms of engagement, such as quotes and replies. Distinct patterns emerge across organisational categories: research facilities, in particular, demonstrate the strongest focus on scholarly tweeting, whereas government accounts are comparatively more successful in eliciting engagement across all metrics, including the more interactive ones. This study contributes both an open dataset of organisational accounts and a methodological framework for their identification, while also highlighting the important roles that organisations play in shaping scholarly discourse on social media.

Er-Te Zheng, Xiaorui Jiang, Zhichao Fang et al.

How academic advantages are transmitted within families is usually studied as occupational inheritance, but it is not clear whether scholarly research orientations persist across generations and if it is an advantage when it does. To address this, we link Wikidata kinship records with OpenAlex bibliometric profiles to study 3,229 documented parent-child scholar pairs and 488,659 publications. Field-level research similarity was evident but not universal: whilst the median similarity was 0.546, 25.3% of parent-child pairs had no Field overlap (i.e., similarity 0). These pairs were substantially more similar than publication-period-matched comparison pairs (median 0.098). Direct academic interaction was uncommon: 10.4% of parent-child pairs had co-authored, 9.8% of children had cited their parents, and 6.9% of parents had cited their children. Nevertheless, each 0.1 increase in Field similarity was associated with 38-39% higher adjusted odds of co-authorship and cross-citation. There was also intergenerational continuity in academic achievement and recognition. Parents' publication volume and field-normalized citation impact were positively associated with those of their children. Children of national academy members had approximately twice the odds of becoming national academy members themselves (Odds Ratio = 2.04), while children of prizewinning parents had 46% higher odds of winning prizes (Odds Ratio = 1.46). However, children of national academy members showed lower research similarity to their parents. Greater research differentiation was associated with higher field-normalized citation impact among children, but not with publication output or higher odds of academic recognition. Academic families therefore appear to transmit resources and advantages with the sole exception that diverging from parental fields seems to confer a citation advantage.

Er-Te Zheng, Hui-Zhen Fu, Mike Thelwall et al.

Timely detection of problematic research is essential for safeguarding scientific integrity. To explore whether social media commentary can serve as an early indicator of potentially problematic articles, this study analysed 3,815 tweets referencing 604 retracted articles and 3,373 tweets referencing 668 comparable non-retracted articles. Tweets critical of the articles were identified through both human annotation and large language models (LLMs). Human annotation revealed that 8.3% of retracted articles were associated with at least one critical tweet prior to retraction, compared to only 1.5% of non-retracted articles, highlighting the potential of tweets as early warning signals of retraction. However, critical tweets identified by LLMs (GPT-4o mini, Gemini 2.0 Flash-Lite, and Claude 3.5 Haiku) only partially aligned with human annotation, suggesting that fully automated monitoring of post-publication discourse should be applied with caution. A human-AI collaborative approach may offer a more reliable and scalable alternative, with human expertise helping to filter out tweets critical of issues unrelated to the research integrity of the articles. Overall, this study provides insights into how social media signals, combined with generative AI technologies, may support efforts to strengthen research integrity.