A Personal Research Agent for Semantic Knowledge Management of Scientific Literature
Bibliographic record
Abstract
The unprecedented rate of scientific publications is a major threat to the productivity of knowledge workers, who rely on scrutinizing the latest scientific discoveries for their daily tasks. Online digital libraries, academic publishing databases and open access repositories grant access to a plethora of information that can overwhelm a researcher, who is looking to obtain fine-grained knowledge relevant for her task at hand. This overload of information has encouraged researchers from various disciplines to look for new approaches in extracting, organizing, and managing knowledge from the immense amount of available literature in ever-growing repositories. \n \nIn this dissertation, we introduce a Personal Research Agent that can help scientists in discovering, reading and learning from scientific documents, primarily in the computer science domain. We demonstrate how a confluence of techniques from the Natural Language Processing and Semantic Web domains can construct a semantically-rich knowledge base, based on an inter-connected graph of scholarly artifacts – effectively transforming scientific literature from written content in isolation, into a queryable web of knowledge, suitable for machine interpretation. \n \nThe challenges of creating an intelligent research agent are manifold: The agent's knowledge base, analogous to his 'brain', must contain accurate information about the knowledge `stored' in documents. It also needs to know about its end-users' tasks and background knowledge. In our work, we present a methodology to extract the rhetorical structure (e.g., claims and contributions) of scholarly documents. We enhance our approach with entity linking techniques that allow us to connect the documents with the Linked Open Data (LOD) cloud, in order to enrich them with additional information from the web of open data. Furthermore, we devise a novel approach for automatic profiling of scholarly users, thereby, enabling the agent to personalize its services, based on a user's background knowledge and interests. We demonstrate how we can automatically create a semantic vector-based representation of the documents and user profiles and utilize them to efficiently detect similar entities in the knowledge base. Finally, as part of our contributions, we present a complete architecture providing an end-to-end workflow for the agent to exploit the opportunities of linking a formal model of scholarly users and scientific publications.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.014 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.003 | 0.002 |
| Scholarly communication | 0.005 | 0.009 |
| Open science | 0.003 | 0.006 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.010 | 0.005 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".