A Pure Visual Approach for Automatically Extracting and Aligning Structured Web Data
Bibliographic record
Abstract
Database-driven websites and the amount of data stored in their databases are growing enormously. Web databases retrieve relevant information in response to users’ queries; the retrieved information is encoded in dynamically generated web pages as structured data records. Identifying and extracting retrieved data records is a fundamental task for many applications, such as competitive intelligence and comparison shopping. This task is challenging due to the complex underlying structure of such web pages and the existence of irrelevant information. Numerous approaches have been introduced to address this problem, but most of them are HTML-dependent solutions that may no longer be functional with the continuous development of HTML. Although a few vision-based techniques have been introduced, various issues exist that inhibit their performance. To overcome this, we propose a novel visual approach, i.e., programming-language-independent, for automatically extracting structured web data. The proposed approach makes full use of the natural human tendency of visual object perception and the Gestalt laws of grouping. The extraction system consists of two tasks: (1) data record extraction, where we apply three of the Gestalt laws (i.e., laws of continuity, proximity, and similarity), which are used to group the adjacently aligned visually similar data records on a web page; and (2) data item extraction and alignment, where we employ the Gestalt law of similarity, which is utilized to group the visually identical data items. Our experiments upon large-scale test sets show that the proposed system is highly effective and outperforms the two state-of-art vision-based approaches, ViDE and rExtractor. The experiments produce an average F1 score of 86.02%, which is approximately 55% and 36% better than that of ViDE and rExtractor for data record extraction, respectively; and an average F1 score of 86.19%, which is approximately 39% better than that of ViDE for data item extraction.
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 | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.003 |
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".