Semantic ETL for Real Time Big Data Analytics
What’s the link between the trends of more and more objects and even commercial transactions on the web being described in a machine-readable, semantic format and the endless streaming of all that data?
Smart Tag
Overview
The web, in particular the read-write web of Social Networking sites is rich in unstructured text. However, most business applications are designed to process quantifiable information, ideally stored in databases. Smart Tag™, First Retail’s Semantic Tagging and Classification Service, is able to automatically tag unstructured content with little or no human intervention and then classify it against available taxonomies.
There are broad uses for such a service ranging from SEO performance improvement and leveraging customer ratings and reviews to the pre-tagging of unstructured text to be combined with other quantitative data.
First Retail’s clients typically have a source of unstructured content and want to add this as a useful data source to already structured or quantifiable content – and that unstructured content is usually tagged by human outsourced or crowd-sourced labor. Automated semantic tagging and classification can be employed with a high degree of precision and without the fatigue-induced inaccuracies that are symptomatic of human workers.
Technology
The Semantic Tag Manager provides complete back-end capability to support the construction of product tag data. This data can be used for real-time Customer Product Tagging interactions and for Business/Admin User Control.
The Semantic Tag Manager handles the consumer tagging workflow, the vendor tag moderation workflow, and business reporting applications.
The consumer tagging workflow system computes the most relevant existing tags. Given a new product input, it suggests likely tags for it based on previous user tags. It also propagates approved tags based on product similarity, and provides semantically-enabled tags (“smart tags”). Optionally it can be used to mark-up products with semantic data such as RDFa, enabling participation within the Linked Open Data community.
The moderation workflow supports the curation of user-applied tags. It uses machine learning components that continually update and extend blacklists based on learning from human curation, and components to support reduction of human curation costs, including such techniques as clustering and batching similar terms for batch curation, use of computational linguistics, and use of golden sources such as WordNet.
The business reporting system provides a set of standard reports over the tags, and exposes RESTful API’s allowing users to retrieve business intelligence related to any aspect of the tagging data. We include semantic software to support a business GUI interface for power users to create custom business reports using a semantic query language such as SPARQL.
Differentiators
- Smart tag suggestions based on semantic similarity among products and among tags
- Intelligent tag propagation based on product similarity with optional rule-based propagation available as an extension
- Next generation tag visualization based on such methods as semantic clustering
- Reduced turn-around time and reduced cost of curation due to application of machine learning and natural language processing
- Ability to extend tags into semantic mark-up for products (e.g. RDFa markup) for increased SEO
To find out how the Smart Tag can help with your business problems contact ustoday.