... For the case of textual data in general, and product descriptions in particular, obtaining ..... more ... For the case of textual data in general, and product descriptions in particular, obtaining ... We exploit this observation to automatically extract high-quality seeds by defining a modified ... supervised learning to improve the per-formance of Naïve Bayes by exploiting large amounts of ...
Proceedings of the 10th international conference on Intelligent user interfaces - IUI '05, 2005
... a shopping assistant around a predicted shop-ping list is a viable approach. Our ongoing work... more ... a shopping assistant around a predicted shop-ping list is a viable approach. Our ongoing work is focused along three dimensions: Improved consumer models and prediction, promotion selection and deployment, and pro-motion planning. The prediction of a shopping list is a ...
Proceedings of the 2004 ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '04, 2004
This paper describes a prototype that predicts the shopping lists for customers in a retail store... more This paper describes a prototype that predicts the shopping lists for customers in a retail store. The shopping list prediction is one aspect of a larger system we have developed for retailers to provide individual and personalized interactions with customers as they navigate through the retail store. Instead of using traditional personalization approaches, such as clustering or segmentation, we learn
We describe an approach to extract attribute-value pairs from product descriptions on the Web. Th... more We describe an approach to extract attribute-value pairs from product descriptions on the Web. The goal is to augment product databases by represent- ing each product as a set of such attribute-value pairs. Such a representation is useful for a variety of tasks where treating the product as a s et of attribute-value pairs is more useful than as an atomic entity. Examples include product recom- mendations, comparison of single products or complete offerings, and demand forecasting. We formulate the extraction as a classificatio n problem and use Na¨ ive Bayes combined with a multi-view semi-supervised algorithm (co-EM). The ex- traction system requires very little initial user supervis ion: using unlabeled data, it automatically extracts an initial seed list that serves a s training data for the clas- sification algorithm. In addition to the automatically extr acted training data, the co-EM algorithm uses the unlabeled data to extract product attributes and values. Finally, the extracted attributes and values are linked to f orm pairs using depen- dency information and co-location scores. We present promising results on Web product descriptions in two categories of sporting goods products.
... For the case of textual data in general, and product descriptions in particular, obtaining ..... more ... For the case of textual data in general, and product descriptions in particular, obtaining ... We exploit this observation to automatically extract high-quality seeds by defining a modified ... supervised learning to improve the per-formance of Naïve Bayes by exploiting large amounts of ...
Proceedings of the 10th international conference on Intelligent user interfaces - IUI '05, 2005
... a shopping assistant around a predicted shop-ping list is a viable approach. Our ongoing work... more ... a shopping assistant around a predicted shop-ping list is a viable approach. Our ongoing work is focused along three dimensions: Improved consumer models and prediction, promotion selection and deployment, and pro-motion planning. The prediction of a shopping list is a ...
Proceedings of the 2004 ACM SIGKDD international conference on Knowledge discovery and data mining - KDD '04, 2004
This paper describes a prototype that predicts the shopping lists for customers in a retail store... more This paper describes a prototype that predicts the shopping lists for customers in a retail store. The shopping list prediction is one aspect of a larger system we have developed for retailers to provide individual and personalized interactions with customers as they navigate through the retail store. Instead of using traditional personalization approaches, such as clustering or segmentation, we learn
We describe an approach to extract attribute-value pairs from product descriptions on the Web. Th... more We describe an approach to extract attribute-value pairs from product descriptions on the Web. The goal is to augment product databases by represent- ing each product as a set of such attribute-value pairs. Such a representation is useful for a variety of tasks where treating the product as a s et of attribute-value pairs is more useful than as an atomic entity. Examples include product recom- mendations, comparison of single products or complete offerings, and demand forecasting. We formulate the extraction as a classificatio n problem and use Na¨ ive Bayes combined with a multi-view semi-supervised algorithm (co-EM). The ex- traction system requires very little initial user supervis ion: using unlabeled data, it automatically extracts an initial seed list that serves a s training data for the clas- sification algorithm. In addition to the automatically extr acted training data, the co-EM algorithm uses the unlabeled data to extract product attributes and values. Finally, the extracted attributes and values are linked to f orm pairs using depen- dency information and co-location scores. We present promising results on Web product descriptions in two categories of sporting goods products.
Uploads
Papers by Marko Krema