In your company, texts are constantly being produced, the text database is growing and growing? Then perhaps you have already made the experience that finding the really important information can be a challenge.
How can a search function for your text database be used so that you can find not only 1:1 matches, but exactly what you were looking for?
Linguistic phenomenon:
This is where Congree Search comes into play. Congree Search is based on linguistic technologies and combines them with statistical methods. Based on the morphosyntactic information obtained about the words in your texts, Congree Search is able to distinguish important words from unimportant words and guarantees that text passages are found even if they contain not only 1:1 matches for the search terms, but also, for example, inflected forms and spelling variants.
Almost 20% of working time is wasted searching for information.
This is mainly caused by the particular shortcomings of the search technology used. Normal search engines work with fuzzy and stemming mechanisms and consider linguistic information only on the surface.
With Congree Search, you obtain more results and improved relevance thanks to linguistic intelligence.
How does linguistic search work?
Linguistically characteristic units such as composites or nominal phrases of adjective and noun are first identified as potential keywords. Since linguistic factors alone are not sufficient to distinguish general language words from specialized words, the potential keywords are then weighted using sophisticated statistical methods.
Intelligent result weighting
according to the Tf-idf measurement
It is generally accepted that keywords that occur frequently in a document are more important than less frequent keywords (term frequency) and that keywords that occur in many documents in the document set are less important than those that occur only in a few documents (inverse document frequency).
Semantic search
An additional factor that is included in the weighting, is the frequency of semantic features that occur in a text.
Congree Search assigns a semantic class to each of the potential keywords identified: agent, thing, instrument, material, process and many more. The relevance of the keywords is statistically determined on the basis of the semantic classes: The more keywords appear per class, the higher they are weighted.
Advanced Match Recognition
Linguistic operations and intelligent matching strategies enable extended search spaces for keywords that are assigned based on terminology:
Transformation
- Network stability vs. network stability
- Reduction of emissions vs. reduction of emissions
Discontinuity
- indexed and mechanical efficiency vs. indexed efficiency
Partial Matching
- Servo drive control vs. drive control
Synonymization
- Reduction of emissions vs. reduction of emissions
Consideration of your terminology
When performing the indexing, we can take your existing company terminology into account. The integration of the terminology guarantees the recognition of relevant keywords and results in your defined technical vocabulary being given a high weighting during a search.
In addition, the knowledge of term relations – such as generic and subordinate terms – stored in the terminology can be used to identify, for example, a rear-wheel drive as a vehicle drive.
We make search successful
Please contact us with your requirements regarding Congree Search. We are happy to provide advice on the various integration options, and will turn searching into finding.
We are also happy to fulfill special requirements regarding search solutions for you.
Your contact persons
Our specialists for intelligent search solutions help with all questions about Congree Search.
E-mail+49 (0) 721 6677570