Hashtag Platform: Machine-assisted multilingual curation engine
“Hashtag Platform” will allow an international Community of Interest or Practice - or even an individual - to set up a site to multilingually curate content from across the web by combining human editorial judgement with "Hashtag Engine", an innovative combination of several advanced language technologies.
Briefly, the system involves:
- Source Curation: Community Members choose sources to be curated (human editorial judgement)
- Semantic Analysis: Hashtag Engine auto-categorises each piece of content published by each source, using a multilingual taxonomy. Individual tags then map each piece to higher-level Themes, chosen by the Community
- Faceted Search & Machine Translation: abstracts of each piece of tagged and Themed content (each linking back to the original source) appear on the site, where an innovative combination of faceted search, federated search and integrated machine translation allows users to quickly find content of interest across language barriers
- Human Validation: Community Members validate/correct the automatic classification, and highlight individual pieces of high-quality content (human editorial judgement), pushing them to the Home Page and out via social media and enewsletters
- Machine learning: the Engine improves its classification accuracy by analysing the Members' editorial judgement: their corrections of its automatically assigned tags, quality judgements, and site usage (each time a user follows a link to the source represents some validation of the tags assigned to it by the Engine).
- Sentiment Analysis and Text Summary: Members can also select multiple pieces of content and extract more value from them via Sentiment Analysis and Text Summary technologies.
Some more details: