Natural Query

NaturalQuery

Knowledge management system (based on free-text search,) sometimes puts limits to data and document retrieve and navigation.

Indeed, appropriate keywords often involve technical lexicons which do not adapt to everyday language. Evodevo has developed semantic search solution in order to bridge the gap between accessibility and knowledge’s resources. Natural Query is based on ontology-based algorithm that ensure a safe and easy search.

Taking advantage of semantic properties defined in ontology design, NQ is thought to retrieve data and documentation without the technical knowledge necessary to its core identification.

Linguistic Integration 

This kind of research does not only lighten the process of finding information, but also expands the usability of the resources by expanding the classes of users that are effectively able to acquire knowledge from it. This is obtained by an ontology-based correlation between selected keyword, nearest correlates and synonyms/hyponyms.

For example, suppose you are looking for information about “flower”. Indeed, a lot of online resources are referred to “flower” without explicitly mentioning the keyword: I could easily find a wiki page about “roses” and “tulips” and other variety. So the research includes both specialized terms, like “roses” (commonly known as hyponyms), and broader terms like “plants” (commonly known as hyponyms). Correlated content is sometimes involved in search processes, so meronimy enable users to find logical related entities, such as petal in the case of flower. Ontology also enables the conceptual modeling, useful to expanding the knowledge and the focus of a research.

In order to avoid overload and redundant information or junk content Evodevo has released a semantic algorithm based on SKOS (Simple Knowledge Organization System) capable of computing semantic distance between user’s query components and related concept in ontology with which documents are organized. By doing this every user is able to easily retrieve information helped by a domain ontology-based algorithm, even if the selected keywords are not grammatically consistent or are used in archaic forms.

 Disambiguation Query
NaturalQuery2

There are a lot of example of polysemic terms (same words with different meanings). Evodevo Natural Query allows to interpret the user’s query, written in natural language, and to get the best of the search results, eliminating automatically any ambiguity.

Faceted search

The free text search engine returns too much resources, making the research very difficult. Some users do not know what is the exact goal of their search and also the technical jargon to search it better.

NaturalQuery3

A faceted search is a technique for accessing information through a classification system, predetermined by a taxonomic order. Every facet corresponds to a dimension useful to describe the resource, choosing the facets (related a particular aspect about the document) it is possible to identify almost univocally the resource.

The resultant documents are ordinated by a ranking, based on the semantic subject and on metadata. The ranking is customizable.

 Linked Open Data

The search results can be represented by ontology’s instances, (developed in accordance with W3C OWL standard; Dublin Core, SKOS etc.) and published as Linked Open Data.

 Integration with search engines

Evodevo Natural Search can be integrated with open source search engines, as Lucene,  Solr and ElasticSearch, and proprietary ones such as Microsoft Fast Search.

 API

Evodevo Natural Query can be interfaced with Java API or with web services which are invoked by applications as .Net, C, C++, PHP etc.

The research is returned in XML, in database and in semantic format Linked Open Data.

 Architecture

Evodevo Natural Query is a web application. Some modules are available in interactive version, or with specific API or with web services.

The product is developed in Java and it is realized for J2EE – Java 2 Enterprise Edition. It is possible to execute it in current operating system (Windows, Linux, Unix, Mac OS etc.) and application server (Apache Tomcat, Oracle Weblogic, IBM Websphere etc.).