Mining and visualising information from RSS feeds: a case study

Recent years have seen “really simple syndication” or “rich site summary”(RSS) syndication of frequently updated content become ubiquitous across the internet. RSS's XML‐based format allows these data to be stored in a semi‐structured format but, despite the presence of online aggregators and readers, and the related work in clustering feeds and mining subjects by keywords, much potentially useful information present in RSS may remain undiscovered. This paper aims to address this issue in an experimental setting.

This paper presents two distinct technologies which employ the semi‐structured nature of RSS content to allow users to mine information directly from raw RSS feeds: occurrence mining counts occurrences of text strings in feeds, whilst value mining mines structured ticker tape numeric data. It describes both technologies and their implementation in an experiment, where 35 students mined small numbers of RSS feeds and visualised the data mined.

This paper analyses the results of the experiment and cites examples of data mined and visualisations produced. The subject matter of data mined is also explored and potential applications of the technologies are considered.

The mining technologies proposed in this paper have been developed to mine textual and numeric data directly from feeds, but can be extended to mine other data types present in RSS and to include other variants like Atom.

These technologies are seen to be applicable to data mining, the role of data and visualisations in social data analysis, issue tracking in news mining and time series analysis.