Bubble Navigator

September 29, 2016 by Dr. Miloslav Konopík, Categories: Faceted Browsers, User Applications

The Bubble Navigator is a tool for visual navigation of semi-structured or semantic web data. Its main strength is the ability to visualize vast amount of data and to provide full-text search in the real-time. When the tool is opened, all data are instantly aggregated (counted) and the user is presented with bubbles representing the count of each document type (article, person, place, location, …). The user can then input a full-text query which is instantly executed and the aggregations are updated. At any given time, the user can press the  + Add filter  button and all the properties of all found resources are aggregated. The size of bubbles again show the count of a given property. E.g. for person document type, the largest bubble is the birth year (it means that most of person documents contain this field), for the birth year the largest bubble is the range 1966-1992. This means that most of the persons in the DBPedia database were born in 1966-1992.

 

All this is possible due to the ElasticSearch tool running in the background. We have indexed the complete DBPedia database from August 2015. The aggregations are configured to deal with string fields, number and date ranges. ElasticSearch provides instant aggregations (counts, statistics) of the data. Full text queries are available all the time. It is even possible to input phrase searches, multi-field searches, fuzzy operators, filed boosting. The text data are analyzed (stemmed, tokenized) and the analyzers are available for many languages. In our demo, we show that the Semantic Web content can be effectively browsed using advanced indexing tools.

The highlights

  • Complete DBPedia knowledge at one glance.

  • Visual overview of the data count distribution.

  • Searching and filtering the data by seeing the visualized results.

  • Full-text search with a rich query language.

  • The ability to select most restrictive property (small bubble) or the property with highest count of results (big bubble).

On-line demo

http://147.228.127.146:9220/

Our team

We are a team of skilled research and professionals that stay on the top of currently extremely attractive fields:

 

  • Natural Language Processing via Machine Learning.

  • Semi-structured or semantic web data search and analysis (we are skilled in the ElasticSearch framework).

  • Modern web application development (Angular.JS / Node.JS stack).

Contact

Dr. Miloslav Konopík, nlp@mail.kiv.zcu.cz

Web pages: http://nlp.kiv.zcu.cz/