Chengran Yang, Ting Zhang, Jinfeng Jiang, Xin Zhou, Haoye Tian, Mingzhe Du, Jieke Shi, Junkai Chen, Yikun Li, Eng Lieh Ouh, Lwin Khin Shar, David Lo. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Real-time collaborative programming is an emerging technology that supports a team of programmers to concurrently view and edit source code documents, with the benefits of enhancing team productivity and reducing project cost. Shared terminal is one crucial component of real-time collaborative programming environments, which facilitates interactive and instant peer support in debugging scenarios. In this study, we propose a novel approach named FlexSecure to address two major challenges in existing shared terminals. FlexSecure supports unconstrained and flexible shared terminal sessions that allow any collaborator to initiate, and meanwhile, preserves the local security of the initiator by incorporating fine-grained permission control to prevent risky command execution. Prototype implementation has validated the feasibility of FlexSecure, and user evaluation has demonstrated its effectiveness and satisfactory performance.

Real-time collaborative programming supports a team of programmers to concurrently edit a shared set of source code at the same time. To support semantic conflict prevention in real-time collaboration, prior work had proposed a dependency-based automatic locking (DAL) approach, which grants locks on selected source code regions based on a set of prefixed rules. To further improve the flexibility of the DAL scheme by utilizing programmers’ knowledge on semantic conflict risks and collaboration requirements, we propose a novel Request-Invitation-Approval (RIA) scheme, which allows any programmer to manually request the editing permission on a locked code region, or invite another programmer to share locks on a region. To support the proposed scheme, we have further proposed two modes for the permission transfer process, and contributed detailed techniques on four request patterns. Prototype system implementation has validated the feasibility of the approach and techniques, and user evaluations have demonstrated the satisfactory usability of the system.

Real-time collaborative programming is a novel approach that enables programmers to simultaneously edit shared source code at the same time, which has been applied in a variety of software development scenarios. To achieve semantic conflict prevention in real-time collaboration, a dependency-based automatic locking (DAL) approach was proposed in prior work, which prevents programmers' concurrent editing on selected source code regions. However, DAL's reliance on source code analysis techniques may lead to failures when there exists syntax errors in the source code. To overcome such limitation, we propose an error-tolerant code segmentation (ECS) approach, as well as supporting algorithms, to improve semantic conflict prevention. Technically, the ECS approach continuously identifies source code regions and maintains their range information to ensure stable source code segmentation and code region tracking during the collaboration process, without the need to deal with complex syntax issues. The proposed approach and algorithms have been implemented in a prototype, and experimental evaluations have demonstrated their effectiveness and efficiency.

Real-time collaborative programming environments support a group of programmers, who are geographically distributed, to concurrently view and edit a shared set of source code in a real-time fashion. However, this emerging technology has not been widely applied yet, and one critical challenge is the semantic conflict during real-time collaboration. Existing conflict prevention approaches are limited in supporting individual source code files only, with pre-programmed rules that might be inflexible. To address these challenges, we propose a novel semantic conflict prevention approach based on source code annotations, which allow programmers to manually, conveniently and flexibly apply fine-grained conflict prevention rules during real-time collaboration. The proposed approach and techniques have been successfully implemented in a prototype system. User studies and performance evaluations have demonstrated the technical feasibility of the proposed approach and techniques, as well as the prototype's satisfactory efficiency and usability.

Real-time collaborative programming supports a team of programmers to concurrently view and edit the same set of source code at the same time, which is beneficial in meeting particular collaboration needs. However, during the collaboration process, programmers are not able to compile and debug the source code with syntactic errors as it is being continuously edited by other collaborators. To address this challenge, we propose a novel approach named Reversion of Error-free Code with Workspace Isolation (RECON) and contribute supporting techniques with prototype implementation. In this approach, the system continuously monitors the files being collaboratively edited, detects source code without syntactic error, and maintains additional Error-free source code copies. Whenever a programmer attempts to compile and debug the code, the system creates a temporary isolated workspace and replaces the source code files with the latest Error-free copies. The proposed approach and solution have been implemented in a prototype system named CoIDEA, which has indicated the feasibility of the scheme and techniques.

Crowd counting plays an increasingly important role in public security. Recently, many crowd counting methods for a single image have been proposed but few studies have focused on using temporal information from image sequences of videos to improve prediction performance. In the existing methods using videos for crowd estimation, temporal features and spatial features are modeled jointly for the prediction, which makes the model less efficient in extracting spatiotemporal features and difficult to improve the performance of predictions. In order to solve these problems, this paper proposes a Parasite-Host Network (PHNet) which is composed of Parasite branch and Host branch to extract temporal features and spatial features respectively. To specifically extract the transform features in the time domain, we propose a novel architecture termed as “Relational Extractor”(RE) which models the multiplicative interaction features of adjacent frames. In addition, the Host branch extracts the spatial features from a current frame which can be replaced with any model that uses a single image for the prediction. We conducted experiments by using our PHNet on four video crowd counting benchmarks: Venice, UCSD, FDST and CrowdFlow. Experimental results show that PHNet achieves superior performance on these four datasets to the state-of-the-art methods.