Alexandros Labrinidis

Estimating the similarity of SQL queries is a fundamental building block for query recommendation systems, workload analysis, and interactive data exploration. We introduce TESS (Tree-Embedding SQL Similarity), a novel query similarity algorithm that constructs a weighted embedding vector from the label tree of a query, capturing both its structural and semantic content. We also present GESS (Global-Embedding SQL Similarity), a lightweight companion algorithm that embeds the full query text as a single vector. Our motivating application is query recommendation for spatiotemporal data exploration, where an analyst’s queries must be compared by semantic intent rather than by syntactic structure alone. We introduce a rigorous evaluation framework and perform a comprehensive evaluation across five datasets and seven metrics. Our experiments show that TESS achieves superior performance across multiple metrics, making it well-suited for interactive recommendation at scale.

Large Language Models (LLMs) are known for their ability to understand and respond to human instructions/prompts. As such, LLMs can be used to produce natural language interfaces for databases. However, LLMs also have an attack surface that, if not properly secured, can cause serious damage. This paper aims to explore the possibilities of exploiting LLMs as an attack surface for SQL injection. The paper proposes a time- and cost-efficient approach to quickly detect malicious prompts by comparing the semantic similarity of the attack against a dedicated list of known patterns.

Traditional recommender systems help users find the most relevant products or services to match their needs and preferences. However, they overlook the preferences of other sides of the market (aka stakeholders) involved in the system. In this paper, we propose to use contextual bandit algorithms in multi-stakeholder platforms where a multi-sided relevance function with adjusting weights is modeled to consider the preferences of all involved stakeholders. Our extensive experimental results on a dataset consisting of MovieLens (1m), IMDB (81k+), and a synthetic dataset show that our proposed approach outperforms the baseline methods and provides a good trade-off between the satisfaction of different stakeholders over time.

Most transit technologies have focused on answering "When will my bus arrive?"; little has been done to answer "How full will my next bus be?" — which also dramatically affects commuters’ quality of life. In this article, we consider the bus fullness problem. We propose two different formulations of the problem, develop multiple predictive models using context-specific features, and evaluate their accuracy using data from the Pittsburgh region. Our predictive models consistently outperform the baselines by up to 8 times.

Data stream management systems (DSMSs) offer the most effective solution for processing data streams by efficiently executing continuous queries (CQs) over the incoming data. CQs inherently have different levels of criticality and hence different levels of expected quality of service (QoS) and quality of data (QoD). We propose DILoS, a novel unified framework that exploits the synergy between priority-based scheduling and load shedding to support differentiated QoS and QoD for multiple classes of CQs. We also propose ALoMa, a general adaptive load manager that automatically tunes the headroom factor, honors the delay target, and is applicable to complex query networks with shared operators.

Big data is transforming all aspects of the human experience. While advances in computing technology lead to exponential increases in data generated and processed, the ability of humans to process data has not changed significantly. We refer to this disparity as the Big Data - Same Humans Problem, and argue that scalability from a human’s point of view — through summarization, ranking, recommendations, outlier detection, personalization, classification, visualization, and crowd-sourcing — is a critical requirement for successful big data systems and applications.

In a broad range of application areas, data is being collected at an unprecedented scale. Decisions that previously were based on guesswork, or on painstakingly handcrafted models of reality, can now be made using data-driven mathematical models. Such Big Data analysis now drives nearly every aspect of society, including mobile services, retail, manufacturing, financial services, life sciences, and physical sciences. Research challenges abound, ranging from heterogeneity of data, inconsistency and incompleteness, timeliness, privacy, visualization, and collaboration, to the tools ecosystem around Big Data.