TY - GEN
T1 - Diversifying relevant search results from social media using community contributed images
AU - Kalakota, Vaibhav
AU - Bansal, Ajay
N1 - Publisher Copyright:
© 2021 IEEE.
PY - 2021/7
Y1 - 2021/7
N2 - Availability of affordable image and video capturing devices as well as rapid development of social networking and content sharing websites has led to the creation of new type of content, Social Media. Any system serving the end user’s query search request should not only take the relevant images into consideration but they also need to be divergent for a well-rounded description of a query. The previous state-of-the-art methods used a number of views of a particular image as one of the key parameters in order to achieve diverse results. This parameter, while improving the overall results, can omit images that are most recently taken. This might not work if the user is interested in recent images when solving the problem of divergence. Secondly, all prior work considered only one of the clustering algorithms for diversification. The performance of most of these clustering techniques is highly data-dependent and might render inefficient for different kinds of datasets. The main focus of this paper is to use visual description of a landmark location by choosing diverse pictures that best describe all the details of a queried location from community-contributed data sets. For this, an end-to-end framework has been built, to retrieve relevant results that are also diverse. Different retrieval re-ranking and diversification strategies are evaluated to find a balance between relevance and diversification. Clustering techniques are employed to improve divergence. A unique fusion approach has been adopted to overcome the dilemma of selecting an appropriate clustering technique and the corresponding parameters, given a dataset to be investigated. Extensive experiments have been conducted on the Flickr Div150Cred dataset. This system has proved to achieve results that are on par with the start-of-art work done on the MediaEval Challenge. This is achieved without using one of the key parameters that contribute to the improved overall metric results - “Number of Views”.
AB - Availability of affordable image and video capturing devices as well as rapid development of social networking and content sharing websites has led to the creation of new type of content, Social Media. Any system serving the end user’s query search request should not only take the relevant images into consideration but they also need to be divergent for a well-rounded description of a query. The previous state-of-the-art methods used a number of views of a particular image as one of the key parameters in order to achieve diverse results. This parameter, while improving the overall results, can omit images that are most recently taken. This might not work if the user is interested in recent images when solving the problem of divergence. Secondly, all prior work considered only one of the clustering algorithms for diversification. The performance of most of these clustering techniques is highly data-dependent and might render inefficient for different kinds of datasets. The main focus of this paper is to use visual description of a landmark location by choosing diverse pictures that best describe all the details of a queried location from community-contributed data sets. For this, an end-to-end framework has been built, to retrieve relevant results that are also diverse. Different retrieval re-ranking and diversification strategies are evaluated to find a balance between relevance and diversification. Clustering techniques are employed to improve divergence. A unique fusion approach has been adopted to overcome the dilemma of selecting an appropriate clustering technique and the corresponding parameters, given a dataset to be investigated. Extensive experiments have been conducted on the Flickr Div150Cred dataset. This system has proved to achieve results that are on par with the start-of-art work done on the MediaEval Challenge. This is achieved without using one of the key parameters that contribute to the improved overall metric results - “Number of Views”.
UR - http://www.scopus.com/inward/record.url?scp=85115846876&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85115846876&partnerID=8YFLogxK
U2 - 10.1109/COMPSAC51774.2021.00060
DO - 10.1109/COMPSAC51774.2021.00060
M3 - Conference contribution
AN - SCOPUS:85115846876
T3 - Proceedings - 2021 IEEE 45th Annual Computers, Software, and Applications Conference, COMPSAC 2021
SP - 376
EP - 385
BT - Proceedings - 2021 IEEE 45th Annual Computers, Software, and Applications Conference, COMPSAC 2021
A2 - Chan, W. K.
A2 - Claycomb, Bill
A2 - Takakura, Hiroki
A2 - Yang, Ji-Jiang
A2 - Teranishi, Yuuichi
A2 - Towey, Dave
A2 - Segura, Sergio
A2 - Shahriar, Hossain
A2 - Reisman, Sorel
A2 - Ahamed, Sheikh Iqbal
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 45th IEEE Annual Computers, Software, and Applications Conference, COMPSAC 2021
Y2 - 12 July 2021 through 16 July 2021
ER -