Professional Search Solutions
Butterfly - federated search and unified information accessButterfly is a search and request oriented portal middleware that allows the portal frontend to abstract from requests to any dibstributed backend system. In addition to distributing requests to the backend systems Butterfly transforms requests according to the query syntax and monitors response behavior. Moreover, depending on any query parameter, Butterfly is able to modify the presentation of backend systems to be queried. By this, Butterfly decouples portal frontend presentation from backend systems that deliver the content. Furthermore, Butterfly also supports rendering of search results and can therefore act as portal frontend for one or more query constellations. Butterfly scales linearly with the number of installed servers and supports sessions in combination with loadbalancers respectively configured.
Butterfly solution bricks
During the designing and developing process of Butterfly ESEMOS has created several "bricks" for special applications. Due to the name "Butterfly" these bricks have got (mostly Greek) names of butterflies:
- Ariadne: search suggestions
- Brenthis: framework for dynamic landing pages
- Cynthia: statistics
- Danaus: dashboard
- Efeus: feeding interface for indices
- Fulminea: trend analysis tool
- Hypnia: Dictionary management tool
- Idaea: search for URL typos
Search engines for intranet search, internet search plus search and navigation for ecommerce and online shops
ESEMOS assists in designing, implementing, and running own search engines.
- Microsoft FAST technolgy: search, navigation, and search suggestions based on FAST ESP 5.x
- Apache Solr and Nutch: search, navigation, and Butterfly-internal whitelist handling for intelligent searches
- Multiphone method: own algorithm for comparing search requests with the list of all known entries (i.e. music artists in music search)
- improves known Soundex and Metaphone algorithms particularly with regard to recognition of names that are not coming from English-speaking areas
- suggestions presented already while entering search queries
- daily processing of hundreds of thousands of search requests
- response times of less than 10 milliseconds
Brenthis: dynamic landing pages
A search engine-friendly internet presence corresponds with a consistent realization of a customer-oriented information architecture. For this we offer:
Efeus: Search ETL framework
Extracting, transforming, and loading data for search (ETL); data feeding from multiple channels (i.e. TV programs, movie/product/actor/director/artist data); used for information enrichment
Idaea: Search for URL typos
Highly performant and intelligent processing of URL typos with subsequent correction resp. web search. The antecedent URL breaker splits URLs and compares the identified tokens to dictionary entries. As a result two or more keywords are delivered that represent the URL content.
Seamless integration of search and online marketing providers
Decision layer
A highly performant solution for preprocessing search requests. Those requests are checked for their affiliation with categories; subsequently, requests are enhanced by their respectice categories. Example: The search query 'madonna' is relevant for the categories music, encyclopedia, and news. The initial search query will therefore be enhanced and expanded to those three search areas.
A search engine-friendly internet presence corresponds with a consistent realization of a customer-oriented information architecture. For this we offer:
- automatically generated content pages with editorially predefined keyword lists
- search engine friendliness by using short and meaningful URLs
Efeus: Search ETL framework
Extracting, transforming, and loading data for search (ETL); data feeding from multiple channels (i.e. TV programs, movie/product/actor/director/artist data); used for information enrichment
Idaea: Search for URL typos
Highly performant and intelligent processing of URL typos with subsequent correction resp. web search. The antecedent URL breaker splits URLs and compares the identified tokens to dictionary entries. As a result two or more keywords are delivered that represent the URL content.
Seamless integration of search and online marketing providers
- Yahoo
- EBay
- Bing
- Autonomy
- Q-Go
Decision layer
A highly performant solution for preprocessing search requests. Those requests are checked for their affiliation with categories; subsequently, requests are enhanced by their respectice categories. Example: The search query 'madonna' is relevant for the categories music, encyclopedia, and news. The initial search query will therefore be enhanced and expanded to those three search areas.