Abstract

Media aggregators combine the convenience of the RSS information feeds with the capabilities of media sharing and distribution. Personal media aggregators not only provide a way to create, store and distribute thematic content but also allow users to customize the way material is collected, transformed and further distributed. We present the architecture and capabilities of the LoaMA personal media aggregator: a tool that handles multimedia contents of various types and provides a mechanism that allows users to compose transformers to manipulate RSS feeds. Transformations can be used to create customized playlists, which in turn can be shared among users as feeds.

1. Introduction

RSS feeds [1] have brought to the Web an alternative way to access information: besides browsing or searching for information, RSS now delivers updates from a subscription list directly to the RSS readers or aggregators of millions of Internet users. Tools like MovableType and Blogger have made publishing easy. Web-based services like Flickr and del.icio.us have enhanced the publishing and sharing process.

RSS techniques have recently been combined with media sharing techniques, leading to podcasting, the practice of making audio files available online in a way that allows software to automatically detect new files and download them.

Users appreciate the benefits of having material of their interest immediately available rather than having to wait for lengthy downloads to complete. As disk capacities and network bandwidth increase, this practice of distribution will achieve wider and wider acceptance.

The ease of reading, though, has lagged the progress in publishing. A number of specialized RSS readers and aggregators have been developed, which simplify subscribing to feeds and allow aggregating several news sources. Next generation browsers hopefully will incorporate the ability to subscribe and read RSS feeds.

The next step in the evolution of news aggregators is the Personal Media Aggregator (PMA), which adds facilities that allow each user to customize the way he deals with the media he collects or distributes.

A PMA is a self-contained mini-application that offers ways to access, collect, filter, select, enjoy and exchange a multiplicity of information and communications organized through the metaphor of channels. Channels are created around a particular topic of interest (buzz, work of art, people, news item, event, anything people wish to discuss about) and allow conveying ideas, opinions, commentaries as well as references to materials in various media.

A PMA enables one-to-many, top-down information distribution as well as bottom-up, one-to-one and many-to-many communication and interaction channels.

A PMA allows multiple levels of customization allowing maximum personalization for every end-user.

To clarify the role of a PMA, let us consider three levels in content distribution: the content itself (Web pages, news articles, blogs, media files); metadata describing the content in the forrn of RSS channels; content aggregations (newspapers, online magazines, blog reviews). Authoring tools are used for creating the content and as mentioned some of them provide means to produce automatically metadata in the form of RSS feeds. Creating aggregations still requires significant human effort in discovering, analyzing, filtering, selecting among a large number of sources and producing a personalized view of such contents. This is quite valuable activity and there are people that spend several hours a day scanning through a large set of sources and preparing their personal account of what they consider most interesting. This activity is done either for hobby or as a professional job. Typically each expert reviewer or site specializes on a certain subject: technology, politics, sports, etc. Alternatively, there are services on the Web that provide aggregated views of news feeds built by automated tools, like Google News [2] or Meerkat Open News Wire service [3].

A PMA provides means to automate the process of generating aggregators that fit the needs and interests of an individual user. Besides allowing the user to select which sources to scan, he can determine how to automatically manipulate the contents coming from those sources. For instance, often sources are typically monothematic, while a user may have several interests. A PMA allows a user to merge multiple sources, classify them according to his own categories of interest and conveniently cluster them in order to discover overlapping material. In fact, as it is common in news media, often several sources will carry the same news item, with only minor changes in presentation.

The capability of manipulating feeds is a distinguishing feature of PMA’s: this might be accomplished by allowing users to parameterize the set of built-in tools, or to program new tools to be added as plug-in’s. The approach that we will describe is based on providing a visual composition tool that allows a user to assemble his own transformations by combing a set of primitive feed transformers, which include classification and clustering.

Besides producing aggregations suiting individual interests, a PMA goes beyond the benefits of Web aggregator services, since it also takes care of fetching the selected material, making it immediately available for reading by the user.

In summary, a PMA is a new way to create, store and distribute thematic content.

In this paper we present LoaMA, a personal media aggregator that has been developed by the students of the course “Laboratorio orientato alle applicazioni” (LOA), for the degree “Laurea Specialistica in Tecnologie Informatiche” at the University of Pisa.

Each year the students of the course select a novel and complex application, whose implementation involves a number of advanced software technologies. The application is designed and implemented through team work and it constitutes the final term project required for the grading of the course.

In the previous years the students developed LoaCMS [7], a Content Management System and Loacker [8], an e-learning platform. Both projects produced fully functional prototypes that have been made available as open source projects on SourceForge and Freshmeat.