Social Network Supported Process Recommender System

Yanming Ye, Jianwei Yin, Yueshen XuThe Scientific World JOURNAL 2014 · DOI

TL;DR

This paper introduces a novel approach to business process recommendation by integrating social network analysis (SNA) with traditional process modeling. By leveraging social features like user interactions and modeling history, their system aims to provide more contextually relevant process suggestions, moving beyond purely structural analysis. This work offers a new perspective on how social dynamics can enhance intelligent business process management.

What Problem Does This Paper Address?

Business process modeling, while crucial for organizational efficiency, is often complex and time-consuming. Existing recommendation systems primarily rely on analyzing the static structure of processes, neglecting valuable social context. This context includes how users interact with processes, their intentions, and their modeling habits. The authors argue that incorporating these social features, akin to how social networks function, can significantly improve the accuracy and relevance of process recommendations, making modeling more efficient and effective.

Methods

The core of this paper is the development of a "social network system of processes." Instead of relying on traditional process mining and retrieval, they propose constructing a social network where processes and their attributes are represented as nodes and relationships. Key methodological contributions include:

  • Social Network Construction: They propose a unique social network model that incorporates more social features than previous attempts, moving beyond simple structural overlaps or user-process matrices.
  • Process Code Definition: A standardized Breadth-First Search (BFS) code is introduced to uniquely represent each process, enabling consistent comparison.
  • Process Matching Degrees: Three distinct metrics are proposed to quantify the similarity between processes:
  • User Matching Degree: This metric combines creator, performer, and reuser similarities, reflecting the collaborative nature of process modeling.
  • Structural Matching Degree: This metric leverages metapaths (sequences of roles and relationships) derived from a social network schema and uses string edit distance to compare them.
  • Behavioral Matching Degree: This metric focuses on the dynamic aspects of processes, comparing functions and events.
  • System Implementation: A prototype system is described, integrating a simple social network site, a query interface, and recommender components that utilize the defined matching degrees.

Datasets

The paper describes experiments conducted on a synthetic dataset generated by a custom generator. This generator is capable of producing processes with various structures (AND-split, OR-split, etc.). The dataset consists of 500 processes, each with randomly assigned activities, performers, and functions. The authors note that this synthetic dataset is similar to those used in prior work but enhanced to include diverse process structures.

Key Insights and Ideas

  • Social Context is Key: The paper strongly advocates that social interactions and user behavior are as important, if not more so, than structural analysis for effective process recommendation.
  • Unified Social Network: They propose a single, richer social network model that can encapsulate information previously spread across multiple, less comprehensive social network types.
  • Metapaths for Structure: The concept of metapaths, derived from a social network schema, offers a novel way to capture structural similarities by representing the flow of information and roles.
  • Combined Matching: The combination of user, structural, and behavioral matching degrees provides a holistic view of process similarity, leading to more nuanced recommendations.

Key Results

The experimental evaluation demonstrates the system's efficiency. While specific quantitative improvements over existing methods are not explicitly detailed in terms of a direct comparison, the results show that the system can provide recommendations within reasonable execution times. Table 2 indicates that varying thresholds for the matching degrees can significantly impact the average execution time and the number of recommended processes, highlighting the configurability of the system. For instance, decreasing the structural matching degree threshold led to increased execution time, suggesting a trade-off between recommendation breadth and computational cost.

Limitations and Future Work

The authors acknowledge several areas for future research:

  • User Community Interests: The importance of considering user communities with similar interests for modeling intentions is highlighted.
  • Process Mining Integration: They suggest that process mining techniques could benefit from the insights gained from social network analysis.
  • More Efficient Methods: The paper calls for the development of more efficient and effective methods and technologies in this emerging field.
  • Parameter Tuning: The experimental results suggest that parameter tuning (thresholds for matching degrees) is crucial for achieving desired performance and recommendation outcomes, implying a need for further investigation into optimal configurations.

Summary updated

Abstract

Process recommendation technologies have gained more and more attention in the field of intelligent business process modeling to assist the process modeling. However, most of the existing technologies only use the process structure analysis and do not take the social features of processes into account, while the process modeling is complex and comprehensive in most situations. This paper studies the feasibility of social network research technologies on process recommendation and builds a social network system of processes based on the features similarities. Then, three process matching degree measurements are presented and the system implementation is discussed subsequently. Finally, experimental evaluations and future works are introduced.
BibTeX
@article{ye2014social,
  author = {Yanming Ye and Jianwei Yin and Yueshen Xu},
  title = {{Social Network Supported Process Recommender System}},
  journal = {The Scientific World JOURNAL},
  year = {2014},
  doi = {10.1155/2014/349065},
}
Figure