An analysis of the abstracts presented at the annual meetings of the Society for Neuroscience from 2001 to 2006

We extracted and processed abstract data from the SFN annual meeting abstracts during the period 2001-2006, using techniques and software from natural language processing, database management, and data visualization and analysis. An important first step in the process was the application of data cleaning and disambiguation methods to construct a unified database, since the data were too noisy to be of full utility in the raw form initially available. The resulting co-author graph in 2006, for example, had 39,645 nodes (with an estimated 6% error rate in our disambiguation of similar author names) and 13,979 abstracts, with an average of 1.5 abstracts per author, 4.3 authors per abstract, and 5.96 collaborators per author (including all authors on shared abstracts). Recent work in related areas has focused on reputational indices such as highly cited papers or scientists and journal impact factors, and to a lesser extent on creating visual maps of the knowledge space. In contrast, there has been relatively less work on the demographics and community structure, the dynamics of the field over time to examine major research trends and the structure of the sources of research funding. In this paper we examined each of these areas in order to gain an objective overview of contemporary neuroscience. Some interesting findings include a high geographical concentration of neuroscience research in north eastern United States, a surprisingly large transient population (60% of the authors appear in only one out of the six studied years), the central role played by the study of neurodegenerative disorders in the neuroscience community structure, and an apparent growth of behavioral/systems neuroscience with a corresponding shrinkage of cellular/molecular neuroscience over the six year period.

💡 Research Summary

This paper presents a comprehensive bibliometric and network analysis of the Society for Neuroscience (SFN) annual meeting abstracts spanning the years 2001 to 2006. The authors collected 13,979 abstracts (13,979 in 2006 alone) and initially identified 39,645 author entries. Because raw author data suffered from homonymy, variable use of initials, and typographical errors, a multi‑step disambiguation pipeline was applied. The pipeline combined string similarity, entity matching, and co‑authorship pattern analysis, yielding an estimated 6 % residual error rate and reducing the author count by roughly 35 % (from an upper bound of 197,429 to a cleaned total of 128,553 unique individuals).

A co‑authorship graph was constructed where each vertex represents a distinct author and an undirected edge indicates at least one shared abstract. In 2006 the graph comprised 39,645 vertices and 13,979 abstracts, with an average degree of 5.96 and an average shortest‑path length of about six steps, confirming a “small‑world” structure typical of scientific collaboration networks. The average number of abstracts per author across the six‑year span was 2.93, while the average number of authors per abstract was 4.31, closely matching global trends in life‑science publications.

Geographic analysis extracted city, state/province, and country information from author affiliations. Mapping these data revealed a pronounced concentration of neuroscience activity in the northeastern United States (Boston, New York, Philadelphia, Baltimore/DC), Southern California, Tokyo, London, and Montreal. The distribution remained relatively stable over the six years, with only a temporary rise in Atlanta representation coinciding with the 2006 meeting location.

Demographically, the study uncovered a striking “transient” population: approximately 60 % of authors appeared in only a single year of the six‑year window. These transient contributors are likely undergraduate, graduate, or post‑doctoral researchers who entered and left the field quickly. Conversely, a smaller core of authors (≈10 % of the total) contributed abstracts in three or more years, representing the stable backbone of the community.

Topic modeling was performed using natural‑language‑processing techniques (Latent Dirichlet Allocation) and cross‑validated against the official SFN thematic categories. The dominant topics included neurodegenerative diseases, behavioral/systems neuroscience, and cellular/molecular neuroscience. Over the study period, behavioral/systems neuroscience grew by roughly 12 % in relative share, while cellular/molecular neuroscience declined by about 9 %. Neurodegenerative disease research occupied a central position in the co‑authorship network, acting as a hub that linked many otherwise disparate sub‑communities.

The authors also mapped National Institutes of Health (NIH) funding to the identified topic clusters. Surprisingly, there was no consistent correlation between funding magnitude and topic growth. For instance, neurodegenerative disease clusters received substantial NIH support yet exhibited modest growth, whereas behavioral/systems neuroscience expanded rapidly despite comparatively lower funding levels. This decoupling suggests that researcher interest and policy priorities do not always align with funding allocations.

Limitations acknowledged include the residual author‑disambiguation error, the lack of peer review for abstracts (making quality assessment difficult), and the relatively short six‑year observation window, which may not capture longer‑term dynamics. The paper proposes extending the methodology to longer time spans and integrating other scholarly outputs such as journal articles and patents to refine models of field evolution.

Overall, the study demonstrates that systematic extraction, cleaning, and analysis of conference abstracts can yield valuable insights into the demographic composition, collaborative structure, thematic shifts, and funding landscape of a scientific discipline. These findings are relevant to researchers planning collaborations, institutions assessing strategic positioning, and policymakers seeking evidence‑based guidance for resource allocation in neuroscience.