Bibliographic record
Abstract
Traditional approaches to information extraction implicitly assume that many elements of the task are static — the user’s query, and the description of domain and corpus, for example. We believe that in many real situations, however, this assumption does not hold and it is important to consider how the system could best support interaction with the user when the assumption breaks down. Current goals in the information extraction community are for the system to produce accurate results while being easy to retrain and port to a new domain. We seek to extend current approaches to handle dynamic elements of the problem. “Evolving queries”, discussed in the information retrieval (IR) literature, need to be supported by information extraction (IE) systems; IR and IE are both, after all, tools for gathering information from documents in response to a user query. When a casual user — neither an expert in the use of the system, nor in the domain — engages in any information gathering task, there will be an initial phase of investigation and discovery during which the user becomes familiar with the system, the domain, and the documents in the corpus and the user’s query may change over time or evolve. For example, a user may have a query about terrorist activities, asking for the names of perpetrators and the locations of targets; an interim system output prompts the user to refine the query, redefining terrorist activities as involving only a subset of weapons while generalizing to allow for additional (e.g., government) perpetrators. The query is not the only element that may change over time; certainly the domain evolves as additional documents are processed. As well, when the corpus is very large or dynamic (e.g., the Internet), the corpus itself may be seen as evolving — rules for mapping text patterns to query items that apply at one time or for one portion of the corpus no longer apply for another. To provide more robust support for information extraction in a dynamic environment, we consider such issues as:
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".