We are very happy to announce today our public showcase of the integration of Apache Stanbol into our Social Media Monitoring Tool Twistory.
This integration is the result of Manafactory’s participation in the IKS Early Adopters program and of a fruitful collaboration with INSIDEOUT10 (another IKS Early Adopter) – INSIDEOUT10 previously worked on WordPress (the same application framework Twistory uses) and added the contributed engines (a complete list is described in the validation page) we used for our scope.
Thanks to Apache Stanbol enhancement engines and to the gateway API that was developed for this project Twistory is now able to enrich tweets as well as web pages linked in those tweets with named entities (people, organizations and places) and to keep track of the resource consumption for these semantic enrichments in terms of calls, data processing units and generated traffic (these can be monitored here http://idntik.it/#/reports/calls). Twistory is available as Software as a Service Model (SaaS) therefore is important for us to keep track for each instance of the resource consumption.
Adding the semantic layer to Twistory at this stage provides users (Social Media Specialists) with a new way to organize the information on the system (tweets and links are now accessible by looking at people, organization and places) and to discover emergent patterns in the conversations (for instance the impact of the upcoming italian political elections in the links can be seen when looking at the list of the organizations).
Here we present a screencast of the trial we’re performing with ATAC SpA (one of our client – the Public Transport Agency for the city of Rome) that describes the semantic features of Twistory.
For this use-case we’re including in the monitoring a set of keyword to filter the incoming twitter firehose (these are: ATAC, MetroA, MetroB) – each tweet is annotated with relevant Freebase/DBpedia entities describing people, organizations and places – moreover for each link contained in these tweets a similar analysis is performed by fetching the URLs and removing any HTML tag with a Stanbol Façade (a custom API front-end for Stanbol that includes Rhino and Readability.js) that was developed for the scope.
The annotations are used to browse tweets and links and are visualized using a set of pie charts.