kSpheres Features

kSpheres empowers the ordinary business user with the ability to achieve results that used to require full time programmers, database architects and big IT budgets
Data Collection & Extraction​


You teach kSpheres how to collect data by pointing and clicking.  It watches and learns, and will then automatically repeat the process hundreds or thousands of times until the job is done -- and then, on a schedule you assign, it will do it all over again, as often as you like.


Machine Learning

The more data kSpheres gathers, and the more often it is used and corrected, the more it learns about how to properly analyse and represent it.  This learning is cumulative: not only will you teach kSpheres, you will benefit from the teaching that thousands of other users have given it.

Semantic Analysis & Ontology


Unlike any other tool, kSpheres uses state-of-the-art technologies that enable it to understand the meaning of the data it is collecting.  This becomes essential when the same thing is called by different names ("surname" and "last name") and when making inferences about the data (if person X is the mother of person Y, person X must be female).


Entity Extraction & Sentiment Analysis

Free form text, meant for human consumption, can also be used by kSpheres to identify the "things" that the text refers to -- products, companies, places, people, brands, events and many other instances of things can be pulled from the text and treated as useful data by kSpheres.
kSpheres can create a Sentiment Analysis (Positive/Negative/Neutral) for extracted entities, product reviews and any free form text.

Data Harmonization & De-Duplication

Most data on the web and in corporate databases is messy.  kSpheres can clean up the mess in one dataset or in thousands, ridding the data of misspelled or anomalous labels ("pounds", "lbs.", "livre" "p") and creating a single, consistent data model about data from many unique sources.


Affinity & Semantic Proximity

Given a word whose meaning is known, kSpheres can identify similar and related concepts, so that, even though an article about an Aston Martin never uses the word "automobile," kSpheres knows it concerns cars, and can propose data that is similar.