Intelligent Software Engineering - API Recommendation - LLM-assisted Developer Tools

Meng Shuai

Member of the PRiSE Research Group, Anhui Polytechnic University

My research direction is Intelligent Software Engineering - API Recommendation. I study how novice programmers express incomplete software-development intents and how intelligent systems can recover those intents to provide reliable API recommendations.

View Research Public CV

Research Focus

Intelligent Software Engineering - API Recommendation

I focus on API recommendation in intelligent software engineering, especially the gap between vague natural-language programming needs and reliable API-level development support.

Academic Training

2024 - Present

M.S. in Software Engineering

Anhui Polytechnic University

Research direction: Intelligent Software Engineering - API Recommendation.

2020 - 2024

B.E. in Data Science and Big Data Technology

Anhui Polytechnic University

Served as president of the Computer Association and participated in programming contests, volunteer service, and student technology activities.

  • Intelligent Software Engineering - API Recommendation
  • API recommendation for novice programmers
  • LLM-assisted software engineering
  • intent clarification and multi-path consensus aggregation
  • developer tools and intelligent programming support
  • IEEE Transactions on Reliability paper published in 2026; second author and corresponding author.
  • CoLAR revised and resubmitted to IEEE Transactions on Services Computing in September 2026; first author.
  • CoHAPI submitted to Automated Software Engineering in April 2026; second author and corresponding author.
  • Hands-on engineering experience in Spring Cloud, Vue, warehouse MES systems, and AI application prototyping.

Publications and Manuscripts

Research Output

My work explores three complementary directions in API recommendation: I/O-example-driven intent clarification, retrieval-guided multi-path consensus aggregation, and learning from heterogeneous information.

01

API Recommendation for Novice Programmers: From Clear Expressions to Effective Results

IEEE Transactions on Reliability - Published · 2026

Second author and corresponding author

2026 · Vol. 75 · pp. 991–1005 · DOI: 10.1109/TR.2026.3664018

IOCAPI clarifies programming intent through user-confirmed I/O examples. Its Intent Detector, Code Generator, and Task Bridger connect LLM-based clarification with code-model generation to deliver API recommendations and interpretable code examples. Evaluation combines three datasets with a user study covering seven programming tasks.

Paper link

  • API recommendation
  • intent modeling
  • I/O examples
  • LLM reasoning
Framework overview of IOCAPI
IOCAPI framework
02

CoLAR: API Recommendation for Novice Programmers via Retrieval-Guided Multi-Path Consensus Aggregation

IEEE Transactions on Services Computing - Revised and resubmitted · Sep 2026

First author

CoLAR retrieves related Q&A examples to guide multiple reasoning paths, then ranks catalog-resolved API candidates using cross-path recurrence and query–documentation similarity. Consensus is an inference-time ranking signal, with no additional consensus model trained. Evaluation covers BIKER-Dataset, APIBENCH-Q, and ReSO-GitBench; gains vary by dataset and metric, and agreement does not guarantee correctness.

Journal link

  • retrieval-guided generation
  • multi-path reasoning
  • consensus aggregation
  • API ranking
Framework overview of CoLAR
CoLAR framework
03

API Recommendation for Novice Programmers with Consensus Learning using Heterogeneous Information

Automated Software Engineering - Submitted · Apr 2026

Second author and corresponding author

CoHAPI trains source-specialized recommendation branches on Stack Overflow Q&A and GitHub code comments. It combines rank-fusion consensus with a contrastive-learning semantic validator to rerank Java API candidates for incomplete natural-language queries. Evaluation on two public benchmarks examines the value of these complementary information sources.

Journal link

  • heterogeneous information
  • Stack Overflow
  • GitHub comments
  • contrastive reranking
Framework overview of CoHAPI
CoHAPI framework

Engineering Practice

Projects with Research and System Value

The project section emphasizes implementation depth, system responsibility, and how engineering practice supports long-term software-engineering research.

AI Alumni Time Machine preview

Research-driven product prototype - 2026 - Present

AI Alumni Time Machine

An AI-powered alumni service project for university history museums, alumni centers, anniversaries, and graduation scenarios. The system explores youthful portrait generation, campus-scene fusion, historical group-photo retrieval, and personalized commemorative content generation.

Project website

Role: Served as project lead, responsible for product planning, system architecture, core feature development, and team coordination from idea validation to prototype delivery.

Outcome: Built an experience-oriented prototype with public-facing project materials and a demonstrable system flow.

  • AI image generation
  • face retrieval
  • interactive web system
  • product prototyping
Warehouse MES Management System preview

Industry internship project - May 2025 - Nov 2025

Warehouse MES Management System

A warehouse management system for Anhui Meirui'er Filter Co., Ltd., covering inventory management, inbound and outbound workflows, batch tracking, stock alerts, and operational dashboards.

Role: Participated as a core developer, designing and implementing key business modules on both backend microservices and frontend visualization pages.

Outcome: Supported high-frequency daily material-flow operations and improved the digital management capability of warehouse workflows.

  • Spring Cloud
  • Vue 2
  • microservices
  • inventory management
  • data visualization

Selected Honors

Awards and Certificates

Representative awards are shown with selected public certificate previews.