The past decade has seen the rapid development of the online newsroom. News published online are the main outlet of news surpassing traditional printed newspapers. This poses challenges to the production and to the consumption of those news. With those many sources of information available it is important to find ways to cluster and organise the documents if one wants to understand this new system.
Traditional approaches to the problem of clustering documents usually embed the documents in a suitable similarity space. Previous studies have reported on the impact of the similarity measures used for clustering of textual corpora [1]. These similarity measures usually are calculated for bag of words representations of the documents. This makes the final document-word matrix high dimensional. Feature vectors with more than 10,000 dimensions are common and algorithms have severe problems with the high dimensionality of the data.
A novel bio inspired approach to the problem of traversing the news is presented. It finds Hamiltonian cycles over documents published by the newspaper The Guardian. A Second Order Swarm Intelligence algorithm based on Ant Colony Optimisation was developed [2, 3] that uses a negative pheromone to mark unrewarding paths with a “no-entry” signal. This approach follows recent findings of negative pheromone usage in real ants [4].
In this case study the corpus of data is represented as a bipartite relation between documents and keywords entered by the journalists to characterise the news. A new similarity measure between documents is presented based on the Q- analysis description [5, 6, 7] of the simplicial complex formed between documents and keywords. The eccentricity between documents (two simplicies) is then used as a novel measure of similarity between documents.
The results prove that the Second Order Swarm Intelligence algorithm performs better in benchmark problems of the travelling salesman problem, with faster convergence and optimal results. The addition of the negative pheromone as a non-entry signal clearly improved the quality of the solutions. The application of the algorithm to the corpus of news of The Guardian creates a coherent navigation system among the news. This allows the users to navigate the news published during a certain period of time in a semantic sequence instead of a time sequence.
This work as broader application as it can be applied to many cases where the data is mapped to bipartite relations (e.g. protein expressions in cells, sentiment analysis, brand awareness in social media, routing problems), as it highlights the connectivity of the underlying complex system.
More than Just Lines on a Map: Best Practices for U.S Bike Routes
TRAVERSING NEWS WITH ANT COLONY OPTIMISATION & NEGATIVE PHEROMONES
1. TRAVERSING NEWS WITH
ANT COLONY OPTIMISATION &
NEGATIVE PHEROMONES
David Sousa-Rodrigues1, and Vitorino Ramos2
ECCS’14, Lucca, IT
September 22-26, 2014
1 – The Open University, UK - david.rodrigues@open.ac.uk !
2 – LaSEEB, ISR, Technical University of Lisbon – vitorino.ramos@ist.utl.pt !
!
2. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 2 / 25
Motivation: Scotland Referendum
3. Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 3 / 25
Outline
• Context of this work and Related Work
• Newspapers
• Adaptive Networks
• Q-analysis
• Community detection
• Ant Colony Optimisation
• Hybrid Connectivity Based Approaches
• TSP Paths in Q-analysis eccentricity matrices
• Conclusions
4. CONTEXT
Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 4 / 25
5. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 5
Context: newspapers (print)
• UK Circulation
/ 25
6. Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 6 / 25
Context: newspapers (electronic)
Internet traffic Internet overtakes print as news
outlet
7. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 7 / 25
Document analysis
• Categorisation of documents (Supervised)
• Machine learning
• K-neighbours, SVM, NN, …
• Clustering (unsupervised)
• Document navigation
• Sometimes associated with clustering
• Information retrieval
• Topic Modelling (LDA...)
8. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Network Science
• Adaptive Networks
• (interplay of topology dynamics and local dynamics
of networks)
• Community Detection in Graphs
• Clustering nodes of graphs
• Q-analysis (Hypernetworks)
• Topological description of the high dimensionality of
structures.
Negative Pheromones 8 / 25
9. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 9 / 25
Bio-inspired & Artificial Life
• Swarm Intelligence algorithms
• Ant Systems
• Ant Colony Optimisation
• Travelling Salesman Problem
• Anti-pheromone ideas
• subtractive anti-pheromone (SAP)
• 1 pheromone – subtracted from poor solutions
• preferential anti-pheromone (PAP)
• 2 pheromones but to solve bi-criterion optimisation problems
10. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 10 / 25
TRAVERSING NEWS
WITH ANT COLONY OPTIMISATION
AND NEGATIVE PHEROMONES
11. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 11 / 25
Two threads
• An extension to the Travelling Salesman Problem
algorithm ACS.
[Rodrigues et al. 2013]
• Application of Q-analysis eccentricity matrices as distance
matrices in the construction of paths in the directed
TSP problem.
[Rodrigues, 2014]
12. Traversing News with Ant Colony Optimisation and
Negative Pheromones 12 / 25
Travelling Salesman Problem
[Visco, 2005] [Sousa-Rodrigues, 2011]
22-26 September 2014
13. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 13 / 25
2nd Order Swarm Intelligence
• Pharaoh ants (Monomorium pharaonis) deposit a
pheromone as a 'no entry' signal to mark unrewarding
foraging paths.
• Double Pheromone Model on top of traditional ACS.
• Traditional positive reinforcement pheromone
• Use of Negative Pheromone to block bad paths.
[Robinson:2007]
[Dorigo:1996]
14. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 14 / 25
State Transition Rule
15. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 15 / 25
Results – Static problems
16. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 16 / 25
Influence of negative pheromone
17. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 17 / 25
Application to dynamic problems:
recovery patterns
18. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 18 / 25
Application to the News
• Articles are represented by a bipartite graph:
• < Document, Tags >
• Documents are connected by their shared vertices (Tags)
Share a vertex
0 - near
Share an edge
1 - near
Share a triangle
2 - near
[Johnson, 2014]
19. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 19 / 25
Application to the News
20. Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 20 / 25
A navigation of q-related news
• Scotland Referendum
• Dataset from GDELT project [http://www.gdeltproject.org/]
• Filtered for events mentioning Scotland during September
• ~ 700 stories / day mentioning Scotland
21. Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 21 / 25
A navigation of q-related news
22. Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 22 / 25
A navigation of q-related news
23. Traversing News with Ant Colony Optimisation and
22-26 September 2014 Negative Pheromones 23 / 25
CONCLUSIONS
24. 22-26 September 2014 Traversing News with Ant Colony Optimisation and
Negative Pheromones 24 / 25
Conclusions
• Bio-inspired algorithm for TSP problems (adaptable to
many problems) – improves on past ACS algorithm.
• Integration of direct information from the simplex structure
into the algorithm for solution finding.
• Creation of coherent corpus of news stories that can be
read in a non timely order but still being highly related.
25. 17 March 2014 Traversing News with Ant Colony Optimisation and
References
Negative Pheromones 25
26. 22-26 September 2014 Negative Pheromones 26 / 25
References
Traversing News with Ant Colony Optimisation and
• D. Sousa-Rodrigues, J. Louçã, and V. Ramos, “From standard to second-order swarm
intelligence phase-space maps,” in 8th European Conference on Complex Systems (S.
Thurner, ed.), (Vienna, Austria), 9 2011.
• V. Ramos, D. Sousa-Rodrigues, and J. Louçã, “Spatio-temporal dynamics on co-evolved
stigmergy,” in 8th European Conference on Complex Systems (S. Thurner,
ed.), (Vienna, Austria), 9 2011.
• V. Ramos, D. Sousa-Rodrigues, and J. Louçã, “Second order swarm intelligence,” in
HAIS’13. 8th Inter- national Conference on Hybrid Artificial Intelligence Systems (J.-S.
Pan, M. Polycarpou, M. Wózniak, A. Carvalho, H. Quintían, and E. Corchado, eds.),
vol. 8073 of Lecture Notes in Computer Science, pp. 411–420, Salamanca, Spain:
Springer Berlin Heidelberg, 9 2013.
• D. Sousa-Rodrigues, “Identifying news clusters using q-analysis and modularity,” in
Proceedings of the European Conference on Complex Systems 2013 (A. Diaz-Guilera,
A. Arenas, and Á. Corral, eds.), (Barcelona), 9 2013.
• J. Johnson, Hypernetworks in the Science of Complex Systems. IMPERIAL COLLEGE
PRESS, Jan. 2014.