arXiv subjects

Zhihong Shen

Publications and source records attributed to Zhihong Shen.

17 recordsLinked to original sources

arXiv2026

Neither Layer Alone: Epistemic Integrity Requires Hierarchical Joint Design for Long-Running AI Agents

Long-running AI agents fail not only when inference fails or tools are underspecified, but when independently evolving model and harness layers change the semantics of belief, capability, and goal commitments across their boundary - a failure class this paper terms Interface Volatility. This paper argues that Agent Epistemic Integrity (AEI) must be treated as a first-class architectural constraint, achievable only through joint model-harness design organized around an explicit interface contract. The central claim is that the model-harness interface contract is the precondition for joint design; its operational form is a four-level hierarchy - goal validity, action-archetype sequencing, tool-instance selection, and invocation-level failure discrimination - that specifies what the boundary must preserve and what structured outputs the model must return for the contract to hold across levels. This reframes long-running agent design away from flat action loops and toward contract-preserving control over persistent state. Evaluation and training should therefore derive from the contract itself, testing whether belief, tool, and goal commitments hold across session boundaries and independent layer upgrades.

cs.SE↗

arXiv2026

DACP: A Scientific Data Access and Collaboration Protocol

Scientific computing is rapidly entering a data-intensive era. However, existing general-purpose network protocol stacks face limitations in eliminating data silos and improving data accessibility and interoperability, making it difficult to effectively meet the demands of emerging paradigms such as AI4Science. To address these challenges, we propose the Data Access and Collaboration Protocol (DACP). DACP defines the Streaming Data Frame (SDF) as its core data model. Through Unified Resource Identification, columnar stream framing, and a reverse supply mechanism, DACP enables data discovery, in-situ computation, and the streaming return of results across scientific data centers, thereby facilitating efficient cross-domain collaboration. Furthermore, this paper introduces faird, a reference server implementation of DACP. This work provides a viable path for building scalable and collaborative scientific data infrastructures.

cs.NI↗

arXiv2026

Reasoning Fails Where Step Flow Breaks

Large reasoning models (LRMs) that generate long chains of thought now perform well on multi-step math, science, and coding tasks. However, their behavior is still unstable and hard to interpret, and existing analysis tools struggle with such long, structured reasoning traces. We introduce Step-Saliency, which pools attention--gradient scores into step-to-step maps along the question--thinking--summary trajectory. Across several models, Step-Saliency reveals two recurring information-flow failures: Shallow Lock-in, where shallow layers over-focus on the current step and barely use earlier context, and Deep Decay, where deep layers gradually lose saliency on the thinking segment and the summary increasingly attends to itself and the last few steps. Motivated by these patterns, we propose StepFlow, a saliency-inspired test-time intervention that adjusts shallow saliency patterns measured by Step-Saliency via Odds-Equal Bridge and adds a small step-level residual in deep layers via Step Momentum Injection. StepFlow improves accuracy on math, science, and coding tasks across multiple LRMs without retraining, indicating that repairing information flow can recover part of their missing reasoning performance.

cs.AI↗

arXiv2023

Pre-training Multi-task Contrastive Learning Models for Scientific Literature Understanding

Scientific literature understanding tasks have gained significant attention due to their potential to accelerate scientific discovery. Pre-trained language models (LMs) have shown effectiveness in these tasks, especially when tuned via contrastive learning. However, jointly utilizing pre-training data across multiple heterogeneous tasks (e.g., extreme multi-label paper classification, citation prediction, and literature search) remains largely unexplored. To bridge this gap, we propose a multi-task contrastive learning framework, SciMult, with a focus on facilitating common knowledge sharing across different scientific literature understanding tasks while preventing task-specific skills from interfering with each other. To be specific, we explore two techniques -- task-aware specialization and instruction tuning. The former adopts a Mixture-of-Experts Transformer architecture with task-aware sub-layers; the latter prepends task-specific instructions to the input text so as to produce task-aware outputs. Extensive experiments on a comprehensive collection of benchmark datasets verify the effectiveness of our task-aware specialization strategy, where we outperform state-of-the-art scientific pre-trained LMs. Code, datasets, and pre-trained models can be found at https://scimult.github.io/.

cs.CL↗

arXiv2023

S2CTrans: Building a bridge from SPARQL to Cypher

In graph data applications, data is primarily maintained using two models: RDF (Resource Description Framework) and property graph. The property graph model is widely adopted by industry, leading to property graph databases generally outperforming RDF databases in graph traversal query performance. However, users often prefer SPARQL as their query language, as it is the W3C's recommended standard. Consequently, exploring SPARQL-to-Property-Graph-Query-Language translation is crucial for enhancing graph query language interoperability and enabling effective querying of property graphs using SPARQL. Despite the substantial differences in semantic representation and processing logic between SPARQL and property graph query languages like Cypher, this paper demonstrates the feasibility of translating SPARQL to Cypher for graph traversal queries using graph relational algebra. We present the S2CTrans framework, which achieves SPARQL-to-Cypher translation while preserving the original semantics. Experimental results with the Berlin SPARQL Benchmark (BSBM) datasets show that S2CTrans successfully converts most SELECT queries in the SPARQL 1.1 specification into type-safe Cypher statements, maintaining result consistency and improving the efficiency of data querying using SPARQL.

cs.DB↗

arXiv2023

Heavy Flavour Physics and CP Violation at LHCb: a Ten-Year Review

Heavy flavour physics provides excellent opportunities to indirectly search for new physics at very high energy scales and to study hadron properties for deep understanding of the strong interaction. The LHCb experiment has been playing a leading role in the study of heavy flavour physics since the start of the LHC operations about ten years ago, and made a range of high-precision measurements and unexpected discoveries, which may have far-reaching implications on the field of particle physics. This review highlights a selection of the most influential physics results on CP violation, rare decays, and heavy flavour production and spectroscopy obtained by LHCb using the data collected during the first two operation periods of the LHC. The upgrade plan of LHCb and the physics prospects are also briefly discussed.

hep-ex↗

arXiv2022

PandaDB: Understanding Unstructured Data in Graph Database

Unstructured data(e.g., images, videos, PDF files, etc.) contain semantic information, for example, the facial feature of a person and the plate number of a vehicle. There could be semantic relationships between data items which are not explicitly represented. For example, a person's face may appear in two irrelevant photos. Also, much information is represented as structured data(e.g., the person's name and age). End-users prefer to query the semantic information from unstructured data together with structured data based on the potential relationships among them. However, due to the lack of a unified database system for structured and unstructured data, developers have to comprise multiple systems and runtime together to answer these queries. In this work, we build an open-source graph database named PandaDB to consistently manage and query structured and unstructured data. We first introduce a graph data model to manage structured and unstructured data, then propose a new query language to understand the semantics of the unstructured data in the graph. Next, we develop a new cost model and related query optimization techniques to speed up the unstructured data processing pipeline. Finally, we optimize the unstructured data storage and provide the index to speed up the query processing over unstructured data. PandaDB is widely used in industrial applications like FinTech, Knowledge Graph, and Recommendation System. The results show PandaDB can support a large scale of unstructured data query processing in a graph.

cs.DB↗

arXiv2022

Metadata-Induced Contrastive Learning for Zero-Shot Multi-Label Text Classification

Large-scale multi-label text classification (LMTC) aims to associate a document with its relevant labels from a large candidate set. Most existing LMTC approaches rely on massive human-annotated training data, which are often costly to obtain and suffer from a long-tailed label distribution (i.e., many labels occur only a few times in the training set). In this paper, we study LMTC under the zero-shot setting, which does not require any annotated documents with labels and only relies on label surface names and descriptions. To train a classifier that calculates the similarity score between a document and a label, we propose a novel metadata-induced contrastive learning (MICoL) method. Different from previous text-based contrastive learning techniques, MICoL exploits document metadata (e.g., authors, venues, and references of research papers), which are widely available on the Web, to derive similar document-document pairs. Experimental results on two large-scale datasets show that: (1) MICoL significantly outperforms strong zero-shot text classification and contrastive learning baselines; (2) MICoL is on par with the state-of-the-art supervised metadata-aware LMTC method trained on 10K-200K labeled documents; and (3) MICoL tends to predict more infrequent labels than supervised methods, thus alleviates the deteriorated performance on long-tailed labels.

cs.CL↗

arXiv2021

MATCH: Metadata-Aware Text Classification in A Large Hierarchy

Multi-label text classification refers to the problem of assigning each given document its most relevant labels from the label set. Commonly, the metadata of the given documents and the hierarchy of the labels are available in real-world applications. However, most existing studies focus on only modeling the text information, with a few attempts to utilize either metadata or hierarchy signals, but not both of them. In this paper, we bridge the gap by formalizing the problem of metadata-aware text classification in a large label hierarchy (e.g., with tens of thousands of labels). To address this problem, we present the MATCH solution -- an end-to-end framework that leverages both metadata and hierarchy information. To incorporate metadata, we pre-train the embeddings of text and metadata in the same space and also leverage the fully-connected attentions to capture the interrelations between them. To leverage the label hierarchy, we propose different ways to regularize the parameters and output probability of each child label by its parents. Extensive experiments on two massive text datasets with large-scale label hierarchies demonstrate the effectiveness of MATCH over state-of-the-art deep learning baselines.

cs.CL↗

arXiv2021

Domain-Specific Pretraining for Vertical Search: Case Study on Biomedical Literature

Information overload is a prevalent challenge in many high-value domains. A prominent case in point is the explosion of the biomedical literature on COVID-19, which swelled to hundreds of thousands of papers in a matter of months. In general, biomedical literature expands by two papers every minute, totalling over a million new papers every year. Search in the biomedical realm, and many other vertical domains is challenging due to the scarcity of direct supervision from click logs. Self-supervised learning has emerged as a promising direction to overcome the annotation bottleneck. We propose a general approach for vertical search based on domain-specific pretraining and present a case study for the biomedical domain. Despite being substantially simpler and not using any relevance labels for training or development, our method performs comparably or better than the best systems in the official TREC-COVID evaluation, a COVID-related biomedical search competition. Using distributed computing in modern cloud infrastructure, our system can scale to tens of millions of articles on PubMed and has been deployed as Microsoft Biomedical Search, a new search experience for biomedical literature: https://aka.ms/biomedsearch.

cs.IR↗

arXiv2021

Is preprint the future of science? A thirty year journey of online preprint services

Preprint is a version of a scientific paper that is publicly distributed preceding formal peer review. Since the launch of arXiv in 1991, preprints have been increasingly distributed over the Internet as opposed to paper copies. It allows open online access to disseminate the original research within a few days, often at a very low operating cost. This work overviews how preprint has been evolving and impacting the research community over the past thirty years alongside the growth of the Web. In this work, we first report that the number of preprints has exponentially increased 63 times in 30 years, although it only accounts for 4% of research articles. Second, we quantify the benefits that preprints bring to authors: preprints reach an audience 14 months earlier on average and associate with five times more citations compared with a non-preprint counterpart. Last, to address the quality concern of preprints, we discover that 41% of preprints are ultimately published at a peer-reviewed destination, and the published venues are as influential as papers without a preprint version. Additionally, we discuss the unprecedented role of preprints in communicating the latest research data during recent public health emergencies. In conclusion, we provide quantitative evidence to unveil the positive impact of preprints on individual researchers and the community. Preprints make scholarly communication more efficient by disseminating scientific discoveries more rapidly and widely with the aid of Web technologies. The measurements we present in this study can help researchers and policymakers make informed decisions about how to effectively use and responsibly embrace a preprint culture.

cs.DL↗

arXiv2020

CORD-19: The COVID-19 Open Research Dataset

The COVID-19 Open Research Dataset (CORD-19) is a growing resource of scientific papers on COVID-19 and related historical coronavirus research. CORD-19 is designed to facilitate the development of text mining and information retrieval systems over its rich collection of metadata and structured full text papers. Since its release, CORD-19 has been downloaded over 200K times and has served as the basis of many COVID-19 text mining and discovery systems. In this article, we describe the mechanics of dataset construction, highlighting challenges and key design decisions, provide an overview of how CORD-19 has been used, and describe several shared tasks built around the dataset. We hope this resource will continue to bring together the computing community, biomedical experts, and policy makers in the search for effective treatments and management policies for COVID-19.

cs.DL↗

arXiv2020

TaxoExpan: Self-supervised Taxonomy Expansion with Position-Enhanced Graph Neural Network

Taxonomies consist of machine-interpretable semantics and provide valuable knowledge for many web applications. For example, online retailers (e.g., Amazon and eBay) use taxonomies for product recommendation, and web search engines (e.g., Google and Bing) leverage taxonomies to enhance query understanding. Enormous efforts have been made on constructing taxonomies either manually or semi-automatically. However, with the fast-growing volume of web content, existing taxonomies will become outdated and fail to capture emerging knowledge. Therefore, in many applications, dynamic expansions of an existing taxonomy are in great demand. In this paper, we study how to expand an existing taxonomy by adding a set of new concepts. We propose a novel self-supervised framework, named TaxoExpan, which automatically generates a set of pairs from the existing taxonomy as training data. Using such self-supervision data, TaxoExpan learns a model to predict whether a query concept is the direct hyponym of an anchor concept. We develop two innovative techniques in TaxoExpan: (1) a position-enhanced graph neural network that encodes the local structure of an anchor concept in the existing taxonomy, and (2) a noise-robust training objective that enables the learned model to be insensitive to the label noise in the self-supervision data. Extensive experiments on three large-scale datasets from different domains demonstrate both the effectiveness and the efficiency of TaxoExpan for taxonomy expansion.

cs.CL↗

arXiv2019

A Scalable Hybrid Research Paper Recommender System for Microsoft Academic

We present the design and methodology for the large scale hybrid paper recommender system used by Microsoft Academic. The system provides recommendations for approximately 160 million English research papers and patents. Our approach handles incomplete citation information while also alleviating the cold-start problem that often affects other recommender systems. We use the Microsoft Academic Graph (MAG), titles, and available abstracts of research papers to build a recommendation list for all documents, thereby combining co-citation and content based approaches. Tuning system parameters also allows for blending and prioritization of each approach which, in turn, allows us to balance paper novelty versus authority in recommendation results. We evaluate the generated recommendations via a user study of 40 participants, with over 2400 recommendation pairs graded and discuss the quality of the results using P@10 and nDCG scores. We see that there is a strong correlation between participant scores and the similarity rankings produced by our system but that additional focus needs to be put towards improving recommender precision, particularly for content based recommendations. The results of the user survey and associated analysis scripts are made available via GitHub and the recommendations produced by our system are available as part of the MAG on Azure to facilitate further research and light up novel research paper recommendation applications.

cs.DL↗

arXiv2018

A Web-scale system for scientific knowledge exploration

To enable efficient exploration of Web-scale scientific knowledge, it is necessary to organize scientific publications into a hierarchical concept structure. In this work, we present a large-scale system to (1) identify hundreds of thousands of scientific concepts, (2) tag these identified concepts to hundreds of millions of scientific publications by leveraging both text and graph structure, and (3) build a six-level concept hierarchy with a subsumption-based model. The system builds the most comprehensive cross-domain scientific concept ontology published to date, with more than 200 thousand concepts and over one million relationships.

cs.CL↗

arXiv2017

Approximations and Bounds for (n, k) Fork-Join Queues: A Linear Transformation Approach

Compared to basic fork-join queues, a job in (n, k) fork-join queues only needs its k out of all n sub-tasks to be finished. Since (n, k) fork-join queues are prevalent in popular distributed systems, erasure coding based cloud storages, and modern network protocols like multipath routing, estimating the sojourn time of such queues is thus critical for the performance measurement and resource plan of computer clusters. However, the estimating keeps to be a well-known open challenge for years, and only rough bounds for a limited range of load factors have been given. In this paper, we developed a closed-form linear transformation technique for jointly-identical random variables: An order statistic can be represented by a linear combination of maxima. This brand-new technique is then used to transform the sojourn time of non-purging (n, k) fork-join queues into a linear combination of the sojourn times of basic (k, k), (k+1, k+1), ..., (n, n) fork-join queues. Consequently, existing approximations for basic fork-join queues can be bridged to the approximations for non-purging (n, k) fork-join queues. The uncovered approximations are then used to improve the upper bounds for purging (n, k) fork-join queues. Simulation experiments show that this linear transformation approach is practiced well for moderate n and relatively large k.

cs.PF↗

arXiv2017

A Century of Science: Globalization of Scientific Collaborations, Citations, and Innovations

Progress in science has advanced the development of human society across history, with dramatic revolutions shaped by information theory, genetic cloning, and artificial intelligence, among the many scientific achievements produced in the 20th century. However, the way that science advances itself is much less well-understood. In this work, we study the evolution of scientific development over the past century by presenting an anatomy of 89 million digitalized papers published between 1900 and 2015. We find that science has benefited from the shift from individual work to collaborative effort, with over 90% of the world-leading innovations generated by collaborations in this century, nearly four times higher than they were in the 1900s. We discover that rather than the frequent myopic- and self-referencing that was common in the early 20th century, modern scientists instead tend to look for literature further back and farther around. Finally, we also observe the globalization of scientific development from 1900 to 2015, including 25-fold and 7-fold increases in international collaborations and citations, respectively, as well as a dramatic decline in the dominant accumulation of citations by the US, the UK, and Germany, from ~95% to ~50% over the same period. Our discoveries are meant to serve as a starter for exploring the visionary ways in which science has developed throughout the past century, generating insight into and an impact upon the current scientific innovations and funding policies.

cs.DL↗