 | |  | |  | |  | |  | |  | |  | |  |
Our Technologies
Ubiquiti software is based on a unique blend of several approaches – which
include techniques from probability & statistics, cognitive sciences,
quantitative machine learning, natural language processing, and databases.
Our software extracts values and terms (as appropriate) which are
quantitative, categorical, syntactic and semantic from datasets. Extraction
of non-quantitative terms is based on a domain-specific ontology which is
created and provided by Ubiquiti, tailored to the needs of each customer.
Technically, the extraction of terms represents a classification problem where
the number of class labels may be numerous, and where the class labels are
structurally organized (i.e., as in an ontology). The datasets are analyzed
with the numeric, categorical and text data included.
The extracted information forms a set of descriptors for the original dataset,
and since the descriptors are accurate, they represent the key information
from the original dataset. Information extraction, equivalently in our context
the descriptors assignment, transforms the original, perhaps unstructured,
dataset – into equivalent datasets with characteristics of structured data.
After the transformation, datasets become amenable to typical analysis as
available for structured datasets. As such, one may apply many available,
well-established means to identify events, problems, opportunities and trends.
Contact Ubiquiti for the Ubiquiti Technology Overview document.
Also, see Publications for Books, Papers and Reports on and by Ubiquiti Inc.
Ubiquiti software is based on a unique blend of several approaches – which
include techniques from probability & statistics, cognitive sciences,
quantitative machine learning, natural language processing, and databases.
Our software extracts values and terms (as appropriate) which are
quantitative, categorical, syntactic and semantic from datasets. Extraction
of non-quantitative terms is based on a domain-specific ontology which is
created and provided by Ubiquiti, tailored to the needs of each customer.
Technically, the extraction of terms represents a classification problem where
the number of class labels may be numerous, and where the class labels are
structurally organized (i.e., as in an ontology). The datasets are analyzed
with the numeric, categorical and text data included.
The extracted information forms a set of descriptors for the original dataset,
and since the descriptors are accurate, they represent the key information
from the original dataset. Information extraction, equivalently in our context
the descriptors assignment, transforms the original, perhaps unstructured,
dataset – into equivalent datasets with characteristics of structured data.
After the transformation, datasets become amenable to typical analysis as
available for structured datasets. As such, one may apply many available,
well-established means to identify events, problems, opportunities and trends.
Contact Ubiquiti for the Ubiquiti Technology Overview document.
Also, see Publications for Books, Papers and Reports on and by Ubiquiti Inc.
Copyright ©2000-2008 Ubiquiti Inc.