Building a Recommender System using Community Level Social Filtering

Alexandra Balahur, Andrés Montoyo

2008

Abstract

Finding the “perfect” product among the dozens of products available on the market is a difficult task for any person. One has to balance between personal needs and tastes, financial limitations, latest trends and social assessment of products. On the other hand, designing the “perfect” product for a given category of users is a difficult task for any company, involving extensive market studies and complex analysis. This paper presents a method to gather the attributes that make up the “perfect” product within a given category and for a specified community. The system built employing this method can be useful for two purposes: firstly, it can recommend products to a user based on the similarity of the feature attributes that most users in his/her community see as important and positive for the product type and the products the user has to opt from. Secondly, it can be used as a practical feedback for companies as to what is valued and how, for a product, within a certain community. For the moment, we will consider the community level as being the country, and thus we will apply and compare the method proposed for English and Spanish. For each product class, we first automatically extract general feature(characteristics describing any product, such as price, size, and design), for each product we then extract specific features (as picture resolution in the case of a digital camera) and feature attributes (adjectives grading the characteristics, as modern or faddy for design). Further on, we use “social filtering” to automatically assign a polarity (positive or negative) to each of the feature attributes, by using a corpus of “pros and cons”-style customer reviews. Additional feature attributes are classified depending on the previously assigned polarities using Support Vector Machines Sequential Minimal Optimization [1] machine learning with the Normalized Google Distance [2]. Finally, recommendations are made by computing the cosine similarity between the vector representing the “perfect” product and the vectors corresponding to products a user could choose from.

References

  1. Platt, J.: Sequential minimal optimization: A fast algorithm for training support vector machines. Microsoft Research Technical Report MSR-TR-98-14 (1998)
  2. Cilibrasi, D., Vitanyi, P.: Automatic Meaning Discovery Using Google. IEEE Journal of Transactions on Knowledge and Data Engineering (2006)
  3. Liu, B.: Web Data Mining. Exploring Hyperlinks, Contents and Usage Data. First edn. Springer (2007)
  4. Hu, M., Liu, B.: Mining Opinion Features in Customer Reviews. In: Proceedings of Nineteenth National Conference on Artificial Intellgience AAAI-2004. (2004)
  5. Dave, K., Lawrence, S., Pennock, D.: Mining the Peanut Gallery: Opinion Extraction and Semantic Classification of Product Reviews. In: Proceedings of WWW-03. (2003)
  6. Fellbaum(ed.), C.: WordNet: An Electronic Lexical Database. First edn. MIT Press (1999)
  7. Vázquez, S., Montoyo, A., Rigau, G.: Using relevant domains resource for word sense disambiguation. In: Proceedings of the ICAI 2004. (2004)
  8. Liu, H., Singh, P.: ConceptNet: A Practical Commonsense Reasoning Toolkit. BT Technology Journal 22 (2004)
  9. Ferrández, A., Palomar, M., Moreno, L.: An Empirical Approach to Spanish Anaphora Resolution. Machine Translation. Special Issue on Anaphora Resolution In Machine Translation. Special Issue on Anaphora Resolution In Machine Translation (1999) 191-216
  10. Lin, D.: Dependency-based Evaluation of MINIPAR. In: In Workshop on the Evaluation of Parsing Systems. (1998)
  11. Polanyi, L., Zaenen, A.: Exploring attitude and affect in text: Theories and applications. Technical Report SS-04-07 (2004)
Download


Paper Citation


in Harvard Style

Balahur A. and Montoyo A. (2008). Building a Recommender System using Community Level Social Filtering.In Proceedings of the 5th International Workshop on Natural Language Processing and Cognitive Science - Volume 1: NLPCS, (ICEIS 2008) ISBN 978-989-8111-45-6, pages 32-41. DOI: 10.5220/0001733200320041


in Bibtex Style

@conference{nlpcs08,
author={Alexandra Balahur and Andrés Montoyo},
title={Building a Recommender System using Community Level Social Filtering},
booktitle={Proceedings of the 5th International Workshop on Natural Language Processing and Cognitive Science - Volume 1: NLPCS, (ICEIS 2008)},
year={2008},
pages={32-41},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0001733200320041},
isbn={978-989-8111-45-6},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 5th International Workshop on Natural Language Processing and Cognitive Science - Volume 1: NLPCS, (ICEIS 2008)
TI - Building a Recommender System using Community Level Social Filtering
SN - 978-989-8111-45-6
AU - Balahur A.
AU - Montoyo A.
PY - 2008
SP - 32
EP - 41
DO - 10.5220/0001733200320041