Faceted search using Solr and Ontopia
Download as PPTX, PDF10 likes2,269 views
This document discusses integrating faceted search capabilities from Apache Solr with structured semantic data from Ontopia. It describes faceted search and how Solr and Ontopia each address search and data representation. An approach is proposed to index Ontopia data in Solr to enable faceted search over semantic concepts. Examples and demos are referenced to illustrate faceted search interfaces.
![Solr – the data modelAn index contains documentsDocuments have fieldsA field can have multiple values{ “id”: “1234”, “title”: “Structure and Interpretation of Computer Programs”, “authors”: [“Harold Abelson”, “Gerald Jay Sussman”] }](https://image.slidesharecdn.com/bouvetone-faceted-search-091123060541-phpapp01/85/Faceted-search-using-Solr-and-Ontopia-25-320.jpg)
Faceted search using Solr and Ontopia
- 1. Faceted search using Solr and Ontopia2009-11-03Geir Ove Grønmo, grove@bouvet.no
- 2. AgendaShort introductions to Solr and OntopiaWhat is faceted search?An integration of the two – a prototypeDemos
- 3. Apache SolrA search engineimplemented as HTTP service on top of Apache Lucenesearching and indexing (no web-crawling)adds support for faceted search (and more)sharding and replicationdistributed searchexcellent interoperability (i.e not really Java-specific)Next release: Solr 1.4Open source:http://lucene.apache.org/solr/Apache Licence 2.0
- 4. OntopiaA Topic Maps toolkit:data representation, persistence and queryingapplication developmentwritten in JavaNext release: Ontopia 5.1Open source:http://code.google.com/p/ontopia/Apache Licence 2.0
- 5. Where the meat is...Solrfast textual search and faceted search supportOntopiarich semantic data and structured searchUser interface designproviding a useful interface to the user
- 6. But first, what is faceted search?A technique for refining search resultsIntegrates textual search and navigationAllows concept compositionslow + expensive + red + used + cararticle + in english + about salmonpeople + aged 20-30 + SQL expertpunk rock songs + < 1 minute + in norwegian + released 1980-1982Support exploration and learningNever returns zero results
- 8. How is it done?Given a starting setusually all documentsor the result of filling in the search input box...do the following:count the number of hits matching each facet fieldwhich fields to facet on are defined at query time
- 12. An example without faceted search
- 13. Facet typesStandard facetsa list of facet valuesHierarchical facet valuestaxonomy of facet valuesRange/query facetsdatespricesalphabet bucketsintervals (lower and upper bounds)
- 14. Standard facets
- 15. Hierarchical facet valuesNote: the facets can also be hierarchical
- 16. Alphabet buckets
- 17. Range facets
- 18. User interface considerationsSingle selectlinkradio buttonMulti selectcheckboxesDecide on which operator to use: AND/ORwithin a facetbetween facetsHow many facet values to displaygiven limited screen real estateHow to provide intuitive undo operation
- 19. Examples
- 20. ScoringSome types of documents should be ranked higher than othersSolr lets one boost the default score:per documentper fieldThe total score of a documents depends on:the boost and score of the fields adjusted by how relevant a field is relatively to the actual querythe boost of the document
- 21. SortingHow to sort the list of facets?by relevanceHow to sort the values of each facet?by number of hitsalphabeticallyHow to sort the search result?by relevancealphabeticallyby date
- 22. Proposition“Concept composition, using faceted search, and Topic Maps is a perfect match”
- 23. Why not use Ontopia only?You can, but it is not optimizedfor this use caseIt lets you implement faceted searchbut it’ll be too slowThe reasons are:all the expensive processing will have to happen at runtime, and not indexing timeinvolves a lot of traversalrelies on the underlying fulltext search enginesearch has limited cacheability
- 24. Trade-offsConsiderations:Search performanceIndexing performanceConsistencyOntopiano indexing overheadresults always up-to-dateSolrvery fast searchindexing overheadindex must be kept up-to-date regularly
- 25. Solr – the data modelAn index contains documentsDocuments have fieldsA field can have multiple values{ “id”: “1234”, “title”: “Structure and Interpretation of Computer Programs”, “authors”: [“Harold Abelson”, “Gerald Jay Sussman”] }
- 26. Ontopia – the data modelA topic map containstopicsand information about themIdentitiesNamesAssociations to other topicsOccurrences (read: non-association properties)
- 27. Integrating Solr and OntopiaProposed solution:Solr indexes constructed from Ontopia queriesFor each document type create a query that extracts data from the topic map to fields in documentsThen do faceting on selected fieldsUse-case specific schema definitionshould be project specific (to some degree)Perform full index or incremental reindex
- 28. Index rule set
- 35. DemoA prototype for Bergen kommune
- 36. Ideas for the futureFaceted search user-interface in Ontopolycould be made declarativeIncremental reindexingrequires tracking changesusually done with a timestampimplement last-modified field in OntopolyAdd optional fourth column for score boost?a float between 0 and 1Ontopia extensions for interacting with SolrJSP tag librarytolog predicates
- 37. More demosEpicurious: recipe searchhttp://www.epicurious.com/tools/searchresults?search=Flickr photo search with hierarchical facetshttp://people.csail.mit.edu/dfhuynh/projects/hierarchical-facets/test.htmlA collection of faceted navigation examples:http://www.flickr.com/photos/morville/collections/72157603789246885/
- 38. More information3 Quick Design Patterns for Better Faceted Searchhttp://www.thingsontop.com/3-quick-patterns-better-facet-design-889.htmlHow to Make a Faceted Classification and Put It On the Webhttp://www.miskatonic.org/library/facet-web-howto.htmlBook: Faceted Search (Synthesis Lectures on Information Concepts, Retrieval, and Services), Daniel Tunkelang
- 39. ...is easier to find when using faceted search.Structured semantics-rich data...