arXiv subjects

Qunzhi Zhou

Publications and source records attributed to Qunzhi Zhou.

5 recordsLinked to original sources

arXiv2026

Iterating Toward Better Search: A Two-Agent Simulation Framework for Evaluating Agentic Search Architectures in E-Commerce

We present a modular two-agent simulation framework for evaluating conversational shopping assistant architectures. An independent buyer agent, configured with personas, missions, and patience levels, is paired with an interchangeable responder that integrates with a real e-commerce search API. Holding the buyer constant across experiments enables controlled comparison of responder designs on identical scenarios. Using 2011 conversations across 14 persona buckets, we establish four empirical findings. First, rolling-window memory outperforms intent-extraction memory on all quality metrics while being 35% faster per query. Second, illustrating rapid evidence-driven iteration, a systematic failure analysis of a responder version enables targeted fixes that reduce failure and near-failure rates by 62% across the full dataset. Third, swapping the responder LLM backbone from Gemini~2.5 to Llama~3.3~70B costs 0.16--0.45 points despite identical architecture. Finally, we document systematic philosophical disagreement between frontier LLM judges: Gemini rewards process correctness while Claude demands concrete outcomes, despite using the same evaluation prompt.

cs.AI↗

arXiv2026

Improving Search Suggestions for Alphanumeric Queries

Alphanumeric identifiers such as manufacturer part numbers (MPNs), SKUs, and model codes are ubiquitous in e-commerce catalogs and search. These identifiers are sparse, non linguistic, and highly sensitive to tokenization and typographical variation, rendering conventional lexical and embedding based retrieval methods ineffective. We propose a training free, character level retrieval framework that encodes each alphanumeric sequence as a fixed length binary vector. This representation enables efficient similarity computation via Hamming distance and supports nearest neighbor retrieval over large identifier corpora. An optional re-ranking stage using edit distance refines precision while preserving latency guarantees. The method offers a practical and interpretable alternative to learned dense retrieval models, making it suitable for production deployment in search suggestion generation systems. Significant gains in business metrics in the A/B test further prove utility of our approach.

cs.IR↗

arXiv2025

Bridging Modality Gaps in e-Commerce Products via Vision-Language Alignment

Item information, such as titles and attributes, is essential for effective user engagement in e-commerce. However, manual or semi-manual entry of structured item specifics often produces inconsistent quality, errors, and slow turnaround, especially for Customer-to-Customer sellers. Generating accurate descriptions directly from item images offers a promising alternative. Existing retrieval-based solutions address some of these issues but often miss fine-grained visual details and struggle with niche or specialized categories. We propose Optimized Preference-Based AI for Listings (OPAL), a framework for generating schema-compliant, high-quality item descriptions from images using a fine-tuned multimodal large language model (MLLM). OPAL addresses key challenges in multimodal e-commerce applications, including bridging modality gaps and capturing detailed contextual information. It introduces two data refinement methods: MLLM-Assisted Conformity Enhancement, which ensures alignment with structured schema requirements, and LLM-Assisted Contextual Understanding, which improves the capture of nuanced and fine-grained information from visual inputs. OPAL uses visual instruction tuning combined with direct preference optimization to fine-tune the MLLM, reducing hallucinations and improving robustness across different backbone architectures. We evaluate OPAL on real-world e-commerce datasets, showing that it consistently outperforms baseline methods in both description quality and schema completion rates. These results demonstrate that OPAL effectively bridges the gap between visual and textual modalities, delivering richer, more accurate, and more consistent item descriptions. This work advances automated listing optimization and supports scalable, high-quality content generation in e-commerce platforms.

cs.IR↗

arXiv2016

Knowledge-infused and Consistent Complex Event Processing over Real-time and Persistent Streams

Emerging applications in Internet of Things (IoT) and Cyber-Physical Systems (CPS) present novel challenges to Big Data platforms for performing online analytics. Ubiquitous sensors from IoT deployments are able to generate data streams at high velocity, that include information from a variety of domains, and accumulate to large volumes on disk. Complex Event Processing (CEP) is recognized as an important real-time computing paradigm for analyzing continuous data streams. However, existing work on CEP is largely limited to relational query processing, exposing two distinctive gaps for query specification and execution: (1) infusing the relational query model with higher level knowledge semantics, and (2) seamless query evaluation across temporal spaces that span past, present and future events. These allow accessible analytics over data streams having properties from different disciplines, and help span the velocity (real-time) and volume (persistent) dimensions. In this article, we introduce a Knowledge-infused CEP (X-CEP) framework that provides domain-aware knowledge query constructs along with temporal operators that allow end-to-end queries to span across real-time and persistent streams. We translate this query model to efficient query execution over online and offline data streams, proposing several optimizations to mitigate the overheads introduced by evaluating semantic predicates and in accessing high-volume historic data streams. The proposed X-CEP query model and execution approaches are implemented in our prototype semantic CEP engine, SCEPter. We validate our query model using domain-aware CEP queries from a real-world Smart Power Grid application, and experimentally analyze the benefits of our optimizations for executing these queries, using event streams from a campus-microgrid IoT deployment.

cs.DB↗

arXiv2013

On Using Complex Event Processing for Dynamic Demand Response Optimization in Microgrid

Demand-side load reduction is a key benefit of Smart Grids. However, existing demand response optimization (DR) programs fail to effectively leverage the near-realtime information available from smart meters and Building Area Networks to respond dynamically to changing energy use profiles. We investigate the use of semantic Complex Event Processing (CEP) patterns to model and detect dynamic situations in a campus microgrid to facilitate adaptive DR. Our focus is on demand-side management rather than supply-side constraints. Continuous data from information sources like smart meters and building sensors are abstracted as event streams. Event patterns for situations that assist with DR are detected from them. Specifically, we offer a taxonomy of event patterns that can guide operators to define situations of interest and we illustrate its efficacy for DR by applying these patterns on realtime events in the USC Campus microgrid using our CEP framework

cs.DC↗