Finding Needles in Emb(a)dding Haystacks: Legal Document Retrieval via Bagging and SVR Ensembles
Kevin Bönisch, Alexander Mehler · 2025
Read Finding Needles in Emb(a)dding Haystacks: Legal Document Retrieval via Bagging and SVR Ensembles by Kevin Bönisch, Alexander Mehler (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-05018.txt.
Description
Kevin Bönisch and Alexander Mehler present a 2025 paper on finding the right legal documents in a sea of text. Lawyers need the few files that match a query, a needle-in-a-haystack problem. The authors reframe search as many small yes-no hunts and answer them with a voting ensemble of regression models.
Their method pairs document embeddings with support vector regression and bootstrap aggregation, called bagging. It needs no deep learning training at all. On a German legal retrieval data set, the ensemble reached recall of 0.849, beating the baseline scores of 0.803 and 0.829. The authors see room to grow by refining the encodings and tuning the settings. Lightweight ensembles may suit legal teams without big compute budgets.
Catalog record
| CREATOR | Kevin Bönisch, Alexander Mehler |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 25 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-05018