The Best of Both Worlds: Combining Information-Theoretic and Computational Private Information Retrieval for Communication Efficiency
Bibliographic record
Abstract
The goal of Private Information Retrieval (PIR) is the ability to query a database successfully without the operator of the database server discovering which record(s) of the database the querier is interested in. There are two main classes of PIR protocols: those that provide privacy guarantees based on the computational limitations of servers, called computational PIR or CPIR, and those that rely on multiple servers not colluding for privacy, called information-theoretic PIR or IT-PIR. These two classes have different advantages and disadvantages that make them more or less attractive to designers of PIR-enabled privacy enhancing technologies. \n \nWe present a hybrid PIR protocol that combines two PIR protocols: one CPIR protocol and one IT-PIR protocol. Our protocol inherits many positive aspects of both classes and mitigates some of the negative aspects. For example, our hybrid protocol maintains partial privacy when the security assumptions of one of the component protocols is broken, mitigating the privacy loss in such an event. We have implemented our protocol as an extension of the Percy++ library so that it combines a PIR protocol by Aguilar Melchor and Gaborit with one by Goldberg. We show that our hybrid protocol uses less communication than either of these component protocols and that our scheme is particularly beneficial when the number of records in a database is large compared to the size of the records. This situation arises in applications such as TLS certificate verification, anonymous communications systems, private LDAP lookups, and others. \n \nThe server-side computation involved in the PIR protocols that we discuss in this thesis all lend themselves to parallelization. As an extension to the Percy++ library we have implemented parallelized server computation for each of these protocols using both multithreading and distributed computation. We show that using parallelization allows the servers to reduce the latency involved in serving PIR queries.
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 | 0.009 | 0.035 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.001 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.002 | 0.006 |
| Scholarly communication | 0.008 | 0.029 |
| Open science | 0.004 | 0.009 |
| Research integrity | 0.003 | 0.005 |
| Insufficient payload (model declined to judge) | 0.009 | 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".