Sunday , October 1 2023

An Integrated Policy System (IPS) for Supporting Policy Making

Maria MOISE1, Victor POPA2
1 Romanian American University
1B Expozitiei Blvd., Bucharest, 012101, Romania
2 I C I Bucharest
(National Institute for R & D in Informatics)

8-10 Averescu Blvd.
011455 Bucharest 1, Romania

Abstract: This paper describes how policy design process is enabled by multichannel social computing, policy topic extraction, semantic analysis, opinions formation, simulation of agent based models and simulation of global models. An Integrated Policy System for policy design process includes: 1) extracting opinions, 2) summarising opinions, 3) generating scenarios, 4) simulating the scenarios for selecting and publishing the best ones. Opinions can be extracted from text analysis or using model for opinion formation on governmental decisions.

Keywords: Individual models, global models, e-Participation, opinions formation, Fuzzy Cognitive Maps (FCMs), Adaptive Neuro Fuzzy Inference Systems (ANFIS).

>>Full Text
Maria MOISE, Victor POPA, An Integrated Policy System (IPS) for Supporting Policy Making, Studies in Informatics and Control, ISSN 1220-1766, vol. 23 (3), pp. 277-284, 2014.

1. Introduction

Modeling the policy making process is a complex task. The policy making process can be modeled at individual and global level. At individual level, public bodies use agent-based simulation, where each agent represents an individual entity (citizen, company, stake-holder, house holder). The agent behavior is observed and taken into account in the policy making process. At global level, the model takes into account objectives, constraints, impacts on environment, economy and society. The integration of individual and global levels is a big challenge.

An Integrated Policy System (IPS) is a multi-disciplinary approach focusing on integration of global models, individual models and e-Participation model of decision process. The IPS uses simulation techniques, social networks, automatic text analysis and opinion formation methods to facilitate integrated e-governance. The citizens are involved in each step of the policy design process and informed about the impact of envisaged decisions.

The policy design process is enabled by multichannel social computing, policy topic extraction, semantic analysis, opinions formation, simulation of agent based models, simulation of global models. An IPS for policy design process includes:

  • extracting opinions;
  • summarising opinions;
  • generating scenarios, simulating the scenarios for selecting the best scenarios, publishing the best scenarios.

Opinions can be extracted from text analysis or using model for opinion formation on a governmental decisions.

The model for opinion formation on a governmental decisions considers a group of agents (individuals) among whom a process of opinion formation takes place. The opinions of an agent will be influenced by opinions of others. These influences are modeled by weights which the agents put on opinions of other agents.

The IPS approach integrates the individual models, global models and e-Participation models using Machine-Learning component.

The Machine Learning component support:

  • learning FCM weights from data;
  • learning agent rules from data;
  • learning rules for clustering and classification of documents.

Learning weights from data for FCM model uses Adaptive Network Based Fuzzy Inference Systems (ANFIS) and Least Square Error (LSE) method. Combining LSE and ANFIS will be done in three steps:

  • decomposing FCM concept domains in regions using ANFIS approach;
  • computing FCM weights for each region using LSE method;
  • aggregating FCM weights for all regions using ANFIS approach.

The ANFIS approach learns the rules and membership functions from data. The ANFIS is an adaptive network of nodes and directional links. Associated with the network is a learning rule – for example, back propagation rule. The network is learning a relationship between inputs and output.

The majority of policy models relay on agent-based simulation [Gill, 2010] where agents represents parties involved in decision-making process.

In addition to agent-based simulation models (individual level models), policy planning needs a global perspective that faces problem at a global level and should tightly interact with individual level. There are two methods concerning integration of global level models with individual level models:

  • global level models are used for taking decisions and individual level models are used to understand the impact of taken decisions;
  • global level models cooperate with individual level models to take decisions and to assess the impact of decisions.

The second approach is studied in several works as:

  • Future Policy Modelling (FUPOL), it is a FP7 project (piloting in Europe and China) focusing on: land use models, housing models, energy models, immigration models, tourism models etc. The project uses opinion mining, FCM and agent-based approaches. FUPOL does not provide integration solutions;
  • ePolicy project, it is a FP7 project, based on game theory to integrate optimization models with simulation models;
  • [Gavanelli 2012] uses Benders decomposition – he decomposes the policy making problem in two components:
    • Master component that defines objectives and constraints;
    • Detail component that defines the strategic action plan to achieve objectives. The detail component communicates with an agent based simulator for generating training data. objectives. The detail component communicates with an agent based simulator for generating training data. His learning method is based on Regression model.
  • [Deng 2007] integrates simulation with optimization (optimization aids simulation for choosing optimal parameters).

The IPS approach extends above research works by:

  • using causal models and opinions formation models [Akiyoshi, 2009], [Moise, Popa, 2014];
  • adding new learning methods based on combining the optimization method LSE with adaptive fuzzy neural networks (ANFIS);
  • learning methods aid both global and local models.

Compared with ePolicy project, IPS adds a new e-participation model for opinion formation, based on agent simulation approach.


  1. ROBINSON, D. T., D. G. BROWN, D. C. PARKER, P. SCHREINEMACHERS, M. A. JANSSEN, M. HUIGENS, H. WITTMER, N. GOTTS, P. PROMBUROM, E. IRWIN, T. BERGER, F. BATZWEILER, C. BARNAUD, Comparison of Empirical Methods for Building Agent-based Models in Land Use Science. Journal of Land Use Science, vol. 2(1), 2007.
  2. HEYDEBRECK, P., M. KLOFSTEN, L. KRÜGER, F2CAn Innovative Approach to Use Fuzzy Cognitive Maps   (FCM)   for   the   Valuation   of High-Technology Ventures, International Business Information Management Association, IBIMA vol. 2011, 2011, Article ID 483.
  3. GAVANELLI, M., F. RIGUZZI, M. MILANO, D. SOTTARA, A. CANGINI, P. CAGNOLI, An Application of Fuzzy Logic to Strategic Environmental Assessment. In Pirrone, R. and Sorbello, F., editors, AI*IA, volume 6934 of LNCS. Springer, 2011.
  4. GAVANELLI, M., F. RIGUZZI, M. MILANO, P. CAGNOLI, Constraint and Optimization Techniques for Supporting Policy Making. In Computational Intelligent Data Analysis for Sustainable Development, chapter 16. Taylor & Francis 2012.
  5. DENG, G., Simulation-based Optimization. PhD thesis, University of Wisconsin – Madison, 2007.
  6. JAFFAR, J, M. MAHER, Constraint Logic Programming, Journal of Logic Programming volume 19/20, 1994,         pp. 503-581.
  7. MATTHEWS, R. N. GILBERT, A. ROACH, G. POLHILL, N. GOTTS, Agent-based Land-use Models: A Review of Applications. Landscape Ecology, volume 22(10), 2007.
  8. GILBERT, N., Computational Social Science. SAGE, 2010.
  9. BRECK, E., C. YEJIN, C. CLAIRE, Identifying Expressions of Opinion in Context, in Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI-2007). 2007.
  10. CARVALHO, P., L. SARMENTO, J. TEIXEIRA, M. J. SILVA. Liars and Saviors in a Sentiment Annotated Corpus of Comments to Political Debates, in Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: short-papers. 2011
  11. FUPOL,
  12. e-Policy,
  13. AKIYOSHI, M., K. KIMURA, H. OISO, N. KOMODA, Analysis Support System of Open-ended Questionnaires Based on Atypical and Typical Opinions Classification, Studies in Informatics and Control ,Vol. 18, No. 3, 2009.
  14. MOISE, M. V. POPA Methods and Experiments Regarding Learning Fuzzy Cognitive Maps, Journal of Information Systems & Operations Management, Vol. 8, No 1, May 2014.