TYPO3 Enterprise Search with Apache Solr
With the new “solr” extension for TYPO3, AOE media offers its clients a high-performance Enterprise Search Solution, capable of delivering a wide range of features in extreme traffic conditions.
As an Enterprise Web CMS, TYPO3 offers a comprehensive range of “out-of-the-box” functionalities. One of these is the TYPO3 search function “Indexed Search”, which is ideally suited to projects of up to 20,000 pages and documents. However, if the project has to handle far larger quantities of data and needs to provide for more complex search criteria, then “Indexed search” begins to slow significantly under the load and very soon becomes impractical as the search index continues to grow.
The Enterprise Search Solution from AOE media, based on the “solr” extension, was developed over a period of ten months in collaboration with several other agencies. It offers search functionality based on Apache’s robust Solr technology, which is capable of maintaining stable high performance under heavy traffic conditions with very large quantities of data.
In addition AOE media developed a highly flexible Indexer which, similar to Google, can crawl and index other websites. This enables unified search and indexing of external shops and micro-sites, even across several different domains, countries and languages: a feature which also makes this solution interesting for websites that do not use TYPO3.
Features of the Enterprise Solr Search in an overview
Categorized search results, exclusion of indefinite articles, etc.
Assorting search results for properties and classes (facettes)
Auto-suggest during search entry
Completement via auto-suggest and auto-complete
"Did you mean ...?" feature with fuzzy search methods
Your search results in a completely individual layout
Alongside its many Enterprise Search functions, the AOE media solution offers other very interesting search features, such as:
- Facetted search
The search results are displayed sorted by characteristic and class (Facets). These facets represent all the possible classification characteristics for the searched object, which can be expressed in groups, such as color, price, materials, finish, release date etc.
- Fuzzy search (did you mean?)
Fuzzy search not only present exact search matches, but also similar results based on type variants. This means that even if you have spelt or expressed your search entry incorrectly, the engine will succeed in finding what you were looking for.
As an aid to entering search terms, the autosuggest function offers auto-completed options to choose from even as you are typing in.
The stemming function displays all those terms that share the same root, or are related to or feature the actual search term within them. Thus a search for “teenager” might also bring up “teen-wear”, “pre-teen clothing”, “make-up tips for teens” etc.
- Extended indexing
The exclusion of stop-words such as indefinite and definite articles (a, an, the) language-dependent text analysis and synonym-search (“automobile” would also bring up “auto” and “car”) are also integrated into the indexing.
- Search result placing (targeted ranking manipulation)
The placing of particular search matches or of paid content in the search results can be made independent of ranking.
- Convincing relevance evaluation and results ranking
Relevance evaluation according to statistical, mathematical and linguistic methodology provides excellent search-result ranking.