|
|
![]() |
![]() |
Accumo Classifier uses textual similarity to summarize and group information into clusters.
Textual similarity is measured using an innovative combination of
Artificial Intelligence, NLP (Natural Language Processing) and Statistical Analysis.
Accumo Classifier uses NLP (Natural Language Processing) algorithms
to perform phrasal linguistic reduction. This ensures that words and phrases with
the same meaning are reduced to a standard form so that similarities
across multiple documents can be detected.
As no two deployment has the same requirements, the linguistic
reduction employed by Accumo Classifier can
easily be customized to suit your organization's particular needs.
Beginning from the first principles of modern Artificial Intelligence,
Accumo Classifier uses an innovative and proprietary algorithm to resolve a set
of text based data into clusters that have maximal evidence of
clusterhood to achieve high quality classification.
|
| Cluster Titles and Strategies |
|
The quality of cluster titles are crucial to the usability of any clustering
system. To title a cluster, Accumo Classifier accurately deduces a meaningful
phrase from the information contained within the cluster. Furthermore, it
strongly constrains the content of each cluster: each cluster is guaranteed
to contain only content posessing the title phrase. This is invaluable for quickly eliminating
irrelevant clusters.
In contrast, some other clustering systems use thesauri to generate cluster titles.
Such cluster titles may have little to do with the actual content of the data.
There are also other approaches that generate only a small number of cluster titles with more
than one word (i.e. phrases). While this approach may obtain better sounding cluster titles,
the actual effect is that they may not span (cover) the entire data set,
and so will not provide sufficient navigation into all available content. As a result, a user may
miss crucial information.
|
|
