Multi-document summarization of scientific corpora
Bibliographic record
Abstract
In this paper, we investigated four approaches for scientific corpora summarization when only gold-standard keyterms available. MEAD with built-in default vocabulary, MEAD with corpus specific vocabulary extracted by Keyphrase Extraction Algorithm (KEA), LexRank (a state-of-the-art summarization algorithm based on random walk) and W3SS (summarization algorithm based on keyword density) are tested on two Computer Science research paper collections. We use a content evaluation method, pyramid method, instead of the well-known ROUGE metrics since there are no gold-standard summaries available for our data. Evaluations with pyramid method indicates that including a corpus specific vocabulary to the traditional summarization methods improves the performance but not significantly. On the other hand, visual inspection shows us that current content evaluation methods, which use only the gold-standard keyterm information, are not intuitive and focus must turn into better evaluation techniques especially for the multi-document summarization problem. Even though the pyramid method looks for important keyterms in the resulting summaries, it cannot distinguish between a general introductory sentence about the area and a specific sentence on the core idea, if they both contain the same keyterm. Also, our results show that the state of the art summarization method LexRank is not feasible for scientific corpus summarization because of its high computational cost.
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 | 0.003 | 0.012 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.006 | 0.004 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".