The increasing number of television channels, on-demand services and online content, is expected to contribute to a better quality of experience for a costumer of such a service. However, the lack of efficient methods for finding the right content, adapted to personal interests, may lead to a progressive loss of clients. In such a scenario, recommendation systems are seen as a tool that can fill this gap and contribute to the loyalty of users. Multimedia content, namely films and television programmes are usually described using a set of metadata elements that include the title, a genre, the date of production, and the list of directors and actors. This paper provides a deep study on how the use of different metadata elements can contribute to increase the quality of the recommendations suggested. The analysis is conducted using Netflix and Movielens datasets and aspects such as the granularity of the descriptions, the accuracy metric used and the sparsity of the data are taken into account. Comparisons with collaborative approaches are also presented.
BackgroundTagging systems enable new modalities of social communication and opportunities for data mining [1]. By being engaged in the annotation process, humans contribute to index information, and it is likely that they are attracted to retrieve information as well. However, a great part of that user-generated annotations have no quality-control process that guarantees the effectiveness of the data collected. Tags created by single users are typically noisy and selfish, contain misspelled words, miss important keywords and are not linked to specific timecodes [1][2][3]. This limits the usefulness of the tags, especially for efficient access to large assets of video content, and, more concretely, it does not allow the accurate access to exact parts of a video.Techniques grounded on video processing are still mostly not feasible given the computer resources required and the difficulty to develop a reliable and universal system to any type of content. Nevertheless, random access to crucial points of videos is essential for retrieving the best content and data that fills our expectations.Many TV and radio broadcasters accumulate very large archives, with various degrees of granularity of content and available metadata. Some of them are very old, speechless Abstract Efficient access to large scale video assets, may it be our life memories in our hard drive or a broadcaster archive which the company is eager to sell, requires content to be conveniently annotated. Manually annotating video content is, however, an intellectually expensive and time-consuming process. In this paper we argue that crowdsourcing, an approach that relies on a remote task force to perform activities that are costly or time-consuming using traditional methods, is a suitable alternative and we describe a solution based on gamification mechanisms for collaboratively collecting timed metadata. Tags introduced by registered players are validated based on a collaborative scoring mechanism that excludes erratic annotations. Voting mechanisms, enabling users to approve or refuse existing tags, provide an extra guarantee on the quality of the annotations. The sense of community is also created as users may watch the crowd's favourite moments of the video provided by a summarization functionality. The system was tested with a pool of volunteers in order to evaluate the quality of the contributions. The results suggest that crowdsourced annotation can describe objects, persons, places, etc. correctly, as well as be very accurate in time. Viana and Pinto Hum. Cent. Comput. Inf. Sci. (2017) Cent. Comput. Inf. Sci. (2017) 7:13 and do not have metadata associated. Most of cataloguing efforts have been geared towards reuse, to enable programme makers to easily find snippets of content to include in their own programmes. However, creating this metadata is a timely and expensive process, still relying on experts that continuously verify and sort the contents to be published, providing structured information for describing the content as a whole, as wel...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.