Forecasting Group-Level Actions Using Similarity Measures

  • Gerardo I. Simari
  • Damon Earp
  • Maria Vanina Martinez
  • Amy Sliva
  • V. S. Subrahmanian


In real-world settings, and in particular in counterterrorism efforts, there is a constant need for a given reasoning agent to have the means by which to “stay ahead” of certain other agents, such as organizations or individuals who may carry out actions against its interests. In this work, we focus on one such way in which a reasoning agent can do this: forecasting group-level actions. This ability is indeed a useful one to have, and one that is definitely attainable if the right kind of data is available. For example, consider the Minorities at Risk Organizational Behavior (MAROB) dataset [4, 19]. This data tracks the behavior for 118 ethnopolitical organizations in the Middle East and Asia Pacific on a yearly basis from 1980 to 2004. For each year, values have been gathered for about 175 measurable variables for each group in the sample. These variables include strategic conditions such as the tendency to commit bombings and armed attacks, as well as background information about the type of leadership, whether the group is involved in cross border violence, etc. Only a subset—around 43—of the approximately 175 attributes in the data represent strategic actions taken by the group, while the others represent variables relating to the environment or context in which the group functions. This context includes variables about the degree of military and financial support the group gets from foreign nations or the ethnic diaspora, the degree of state government repression and persecution against the group, and so forth. It also includes variables about the structure of the group and how factionalized it may or may not be, the level of violence and protests in which the group engages, and the amount of participation in the political process.


Distance Function Past Behavior Manhattan Distance Behavioral Rule Context Attribute 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.



Some of the authors of this paper were funded in part by AFOSR grant FA95500610405, ARO grant W911NF0910206 and ONR grant N000140910685.


  1. 1.
    Antoniou G, van Harmelen F (2004) A semantic web primer. MIT, Cambridge, MAGoogle Scholar
  2. 2.
    Bond J, Petroff V, O’Brien S, Bond D (2004) Forecasting turmoil in indonesia: an application of hidden markov models. International Studies Association, USA. pp 17–21Google Scholar
  3. 3.
    Bowerman B, O’Connell R, Koehler A Forecasting, Time series and regression, 4th edn. Southwestern College Publishers, Cincinnati, OH, USA (2004)Google Scholar
  4. 4.
    CIDCM (2008) Minorities at risk organizational behavior dataset, minorities at risk project. Retrieved from
  5. 5.
    Fagin R, Halpern JY, Megiddo N (1990) A logic for reasoning about probabilities. Inf Comput 87(1/2):78–128Google Scholar
  6. 6.
    Hailperin T (1984) Probability logic. Notre Dame J Form Log 25(3):198–212Google Scholar
  7. 7.
    Khuller S, Martinez MV, Nau D, Simari GI, Sliva A, Subrahmanian VS (2007) Finding most probable worlds of probabilistic logic programs. In: Proceedings 2007 international conference on scalable uncertainty management. Lecture notes in computer science vol 4772. Springer, Berlin/Heidelberg, pp 45–59Google Scholar
  8. 8.
    Khuller S, Martinez MV, Nau DS, Sliva A, Simari GI, Subrahmanian VS (2007) Computing most probable worlds of action probabilistic logic programs: scalable estimation for 1030, 000 worlds. Ann Math Artif Intell 51(2–4):295–331Google Scholar
  9. 9.
    Nilsson N (1986) Probabilistic logic. Artif Intell 28:71–87Google Scholar
  10. 10.
    Schrodt P (2000) Forecasting conflict in the balkans using hidden markov models. American Political Science Association, USAGoogle Scholar
  11. 11.
    Simari GI, Subrahmanian VS (2010) Abductive inference in probabilistic logic programs. Schloss Dagstuhl - Leibniz-Zentrum fuer Informatik 2010, ISBN 978-3-939897-17-0Google Scholar
  12. 12.
    Simari GI, Sliva A, Nau D, Subrahmanian VS (2006) A stochastic language for modelling opponent agents. In: AAMAS ’06: proceedings of the fifth international joint conference on autonomous agents and multiagent systems. ACM, New York, pp 244–246Google Scholar
  13. 13.
    Simari GI, Dickerson JP, Subrahmanian VS (2010) Cost-based query answering in action probabilistic logic programs. Lecture Notes in Computer Science 6379 Springer 2010, ISBN 978-3-642-15950-3Google Scholar
  14. 14.
    Simari GI, Martinez MV, Sliva A, Subrahmanian VS (2012) Focused most probable world computations in probabilistic logic programs. Ann Math Artif Intell 64(2–3):113–143Google Scholar
  15. 15.
    Sliva A, Martinez MV, Simari GI, Subrahmanian VS (2007) SOMA models of the behaviors of stakeholders in the afghan drug economy: a preliminary report. In: First international conference on computational cultural dynamics (ICCCD 2007). AAAI Press, USAGoogle Scholar
  16. 16.
    Sliva A, Subrahmanian VS, Martinez MV, Simari GI (2009) CAPE: automatically predicting changes in group behavior. In: Mathematical methods in counterterrorism. Springer, Wien, pp 247–263Google Scholar
  17. 17.
    Subrahmanian VS (2007) Cultural modeling in real-time. Science 317(5844):1509–1510Google Scholar
  18. 18.
    Subrahmanian VS, Albanese M, Martinez MV, Nau D, Reforgiato D, Simari GI, Sliva A, Wilkenfeld J (2007) Cara: a cultural adversarial reasoning architecture. IEEE Intell Syst 22(2), 12–16Google Scholar
  19. 19.
    Wilkenfeld J, Asal V, Johnson C, Pate A, Michael M (2007) The use of violence by ethnopolitical organizations in the middle east. Technical report, National Consortium for the Study of Terrorism and Responses to TerrorismGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2013

Authors and Affiliations

  • Gerardo I. Simari
    • 1
  • Damon Earp
    • 2
  • Maria Vanina Martinez
    • 1
  • Amy Sliva
    • 3
  • V. S. Subrahmanian
    • 4
  1. 1.Department of Computer ScienceUniversity of OxfordOxfordUK
  2. 2.University of Maryland Institute for Advanced Computer Studies (UMIACS), University of Maryland College ParkCollege ParkUSA
  3. 3.College of Computer and Information ScienceNortheastern UniversityBostonUSA
  4. 4.Department of Computer ScienceUniversity of Maryland College ParkCollege ParkUSA

Personalised recommendations