Nowadays, Web is characterized by a growing availability of multimedia data together with a strong need for integrating different media and modalities of interaction. Hence, one of the main challenges is to bring into the Web data thought and produced for different media, such as TV or press content. In this scenario, we focus on multimodal news aggregation retrieval and fusion. Multimodality, here, is intended as the capability of processing, gathering, manipulating, and organizing data from multiple media. In particular, we tackle two main issues: to extract relevant keywords to news and news aggregations, and to automatically associate suitable advertisements to aggregated data. To achieve the first goal, we propose a solution based on the adoption of extraction-based text summarization techniques; whereas to achieve the second goal, we developed a contextual advertising system that works on multimodal aggregated data. To assess the proposed solutions, we performed experiments on Italian news aggregations. Results show that, in both cases, the proposed solution performs better than the adopted baseline solutions.
Content-based Keywords Extraction and Automatic Advertisement Associations to Multimodal News Aggregations
ARMANO, GIULIANO;GIULIANI, ALESSANDRO;VARGIU, ELOISA
2013-01-01
Abstract
Nowadays, Web is characterized by a growing availability of multimedia data together with a strong need for integrating different media and modalities of interaction. Hence, one of the main challenges is to bring into the Web data thought and produced for different media, such as TV or press content. In this scenario, we focus on multimodal news aggregation retrieval and fusion. Multimodality, here, is intended as the capability of processing, gathering, manipulating, and organizing data from multiple media. In particular, we tackle two main issues: to extract relevant keywords to news and news aggregations, and to automatically associate suitable advertisements to aggregated data. To achieve the first goal, we propose a solution based on the adoption of extraction-based text summarization techniques; whereas to achieve the second goal, we developed a contextual advertising system that works on multimodal aggregated data. To assess the proposed solutions, we performed experiments on Italian news aggregations. Results show that, in both cases, the proposed solution performs better than the adopted baseline solutions.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
