A Fuzzy-Match Search Engine for Physician Directories
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
BACKGROUND: A search engine to find physicians' information is a basic but crucial function of a health care provider's website. Inefficient search engines, which return no results or incorrect results, can lead to patient frustration and potential customer loss. A search engine that can handle misspellings and spelling variations of names is needed, as the United States (US) has culturally, racially, and ethnically diverse names. OBJECTIVE: The Marshfield Clinic website provides a search engine for users to search for physicians' names. The current search engine provides an auto-completion function, but it requires an exact match. We observed that 26% of all searches yielded no results. The goal was to design a fuzzy-match algorithm to aid users in finding physicians easier and faster. METHODS: Instead of an exact match search, we used a fuzzy algorithm to find similar matches for searched terms. In the algorithm, we solved three types of search engine failures: "Typographic", "Phonetic spelling variation", and "Nickname". To solve these mismatches, we used a customized Levenshtein distance calculation that incorporated Soundex coding and a lookup table of nicknames derived from US census data. RESULTS: Using the "Challenge Data Set of Marshfield Physician Names," we evaluated the accuracy of fuzzy-match engine-top ten (90%) and compared it with exact match (0%), Soundex (24%), Levenshtein distance (59%), and fuzzy-match engine-top one (71%). CONCLUSIONS: We designed, created a reference implementation, and evaluated a fuzzy-match search engine for physician directories. The open-source code is available at the codeplex website and a reference implementation is available for demonstration at the datamarsh website.
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.
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 | 0.006 | 0.003 |
| 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 | 0.001 | 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 it