Extraction of Effective Textual and Semantic Features in Learning to Rank for Web Document Retrieval
Bibliographic record
Abstract
Ranking algorithms, as the core of web search systems, are responsible for finding and ranking the most relevant documents to user information needs from the crawled and indexed corpus. With the ever-increasing amount of available training data, ranking technologies are moving towards using Machine Learning methods, described as Learning to Rank algorithms. The basic Learning to Rank systems mainly have used textual features while ignoring semantic features. With the advent of Semantic Web, there is an emerging interest in developing and using semantic features for Learning to Rank systems. An important challenge is that there is currently no comprehensive study on the combined usage of textual and semantic features for Learning to Rank systems. In this paper, first, we define and implement four new sets of semantic features based on Knowledge Graph, Entity Repetition, Textual Fields and Vector Representation of Words and Texts. For experimental analysis, we used the MQ-2007 dataset from LETOR 4, which includes a set of textual features. The results of running six standard Learning to Rank Algorithms show that by using semantic features, either in isolation or in combination with textual features, significantly increases the performance. The increase in performance is even more significant when we limit the tests to hard queries. We also implemented an existing Feature Selection algorithm to test whether it can improve the results even further. The results showed improvements for some Learning to Rank algorithms, yet failed to improve on others.
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 | 0.001 | 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 | 0.002 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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".