arXiv subjects

Ioanna Tsakalidou

Publications and source records attributed to Ioanna Tsakalidou.

3 recordsLinked to original sources

arXiv2026

Semantic Recall for Vector Search

We introduce Semantic Recall, a novel metric to assess the quality of approximate nearest neighbor search algorithms by considering only semantically relevant objects that are theoretically retrievable via exact nearest neighbor search. Unlike traditional recall, semantic recall does not penalize algorithms for failing to retrieve objects that are semantically irrelevant to the query, even if those objects are among their nearest neighbors. We demonstrate that semantic recall is particularly useful for assessing retrieval quality on queries that have few relevant results among their nearest neighbors-a scenario we uncover to be common within embedding datasets. Additionally, we introduce Tolerant Recall, a proxy metric that approximates semantic recall when semantically relevant objects cannot be identified. We empirically show that our metrics are more effective indicators of retrieval quality, and that optimizing search algorithms for these metrics can lead to improved cost-quality tradeoffs.

cs.IR↗

arXiv2026

QVCache: A Query-Aware Vector Cache

Vector databases have become a cornerstone of modern information retrieval, powering applications in recommendation, search, and retrieval-augmented generation (RAG) pipelines. However, scaling approximate nearest neighbor (ANN) search to high recall under strict latency SLOs remains fundamentally constrained by memory capacity and I/O bandwidth. Disk-based vector search systems suffer severe latency degradation at high accuracy, while fully in-memory solutions incur prohibitive memory costs at billion-scale. Despite the central role of caching in traditional databases, vector search lacks a general query-level caching layer capable of amortizing repeated query work. We present QVCache, the first backend-agnostic, query-level caching system for ANN search with bounded memory footprint. QVCache exploits semantic query repetition by performing similarity-aware caching rather than exact-match lookup. It dynamically learns region-specific distance thresholds using an online learning algorithm, enabling recall-preserving cache hits while bounding lookup latency and memory usage independently of dataset size. QVCache operates as a drop-in layer for existing vector databases. It maintains a megabyte-scale memory footprint and achieves sub-millisecond cache-hit latency, reducing end-to-end query latency by up to 40-1000x when integrated with existing ANN systems. For workloads exhibiting temporal-semantic locality, QVCache substantially reduces latency while preserving recall comparable to the underlying ANN backend, establishing it as a missing but essential caching layer for scalable vector search.

cs.DB↗

arXiv2024

Declarative Concurrent Data Structures

Implementing concurrent data structures is challenging and requires a deep understanding of concurrency concepts and careful design to ensure correctness, performance, and scalability. Further, composing operations on two or more concurrent data structures often requires a synchronization wrapper to ensure the operations are applied together atomically, resulting in serialization and, thereby, giving up the performance benefit of the individual data structures. DBMS provides generalized concurrency control (CC) and is a good fit for implementing concurrent data structures. However, DBMSs are over-generalized for this use case, which fails to match the performance of specialized implementations. This paper makes the case for the Declarative Concurrent Data Structures (DCDS) framework for automatically generating concurrent data structures from a serial specification. In DCDS, users declare the attributes and methods needed for their desired data structure through an embedded DSL at design time. DCDS automatically injects CC at build-time, generating a concurrent intermediate representation (IR) compiled into machine code. A declarative interface for designing data structure enables efficient composability through co-optimizing component structures; optimizations are applied to both the composed serial specification and the generated concurrent IR. We realize the DCDS framework in our prototype system Rosti and experimentally show that data structures declared in Rosti can be efficiently composed by co-optimizing their logical functionality and the generated CC protocol. Our evaluation shows that composing a map and a list to create an LRU container can benefit up to 2X performance scalability in Rosti compared to an open-source library. We demonstrate the applicability of DCDS as an in-process OLTP by comparing it with in-memory DBMS, Proteus, and showing up to 2X performance gains.

cs.DB↗