Pertinence Mining develops several products around its KENiA^® (Knowledge Extraction and Notification Architecture) technology for textual information processing in many languages. Its mission is to supply its customers with technologies allowing them to make the best use of available relevant information.

The KENiA^® architecture is modular, robust and extensible. New information processing applications can benefit from external linguistic resources and natural language processing algorithms.

KENiA ® architecture
Knowledge Extraction and Notification Architecture
 

Pertinence Summarizer performs linguistic processing over a document and evaluates the pertinence (the relevance) of its sentences. The process takes into account not only general and/or specialized linguistic markers depending on the nature of the document analyzed, but also the user’s keywords, and optionally terminological bases, to enhance the relevance of the selected sentences.

Pertinence Summarizer handles more than ten languages, including many West European languages, Asian languages (Japanese, Chinese and Korean), Arabic and Russian.

The document formats supported by Pertinence Summarizer are: HTML, PDF, DOC, RTF, TXT and other formats such as PPT, XLS according to the platform used.

Pertinence Summarizer is accessible with a web browser. It is available on a server or installed on a workstation. Pertinence Summarizer can be installed on any workstation equipped with a Java virtual machine. A Java API and a Web API service also allows you to integrate Pertinence Summarizer’s functionalities into your applications.

Features :

• Multilingual : supports ten languages (West European and Asian languages, Arabic and Russian);
• Multiformat : supports various formats (HTML, PDF, DOC, RTF, TXT and PPT, XLS according to the platforms used);
• Handles XML documents (after automatic conversion);
• Several summary display styles (XSLT style sheets);
• Multi-level highlighting of important sentences;
• Highlighting of concepts of the domain (thesaurus or dictionary) and user keywords with the possibility to look up their occurrences in the text using the Tab key;
• Possibility to hide automatically sentences containing specific keywords to exclude;
• Each sentence in the original text can be hidden separately to personalize the summary;
• Retrieve context to reestablish consistence;
• The reduction rate of the summary can be typed in or selected from a list;
• The edited summary can be extracted from the document and copied to the clipboard;
• Available in two versions: server and workstation. Java API and Web API services are available;
• Deployment on any system equipped with a Java virtual machine (Window, Linux, Unix).

Pertinence Meta Search permits the user to submit periodically queries to search engines. Resulting documents are then analyzed and filtered by using relevancy operators. Only really relevant documents are finally retained. Rejected documents are placed in a dedicated folder so that the user can check the filtering process.

Features :

• Multilingual : supports ten languages (West European and Asian languages, Arabic and Russian);
• Multiformat : supports various formats (HTML, PDF, DOC, RTF, TXT and PPT, XLS according to the platforms used);
• Includes the automatic summarization engine Pertinence Summarizer with its complete features described just above;
• Can take into account letter case (eg. bush vs. Bush, roche vs. Roche vs. ROCHE);
• Can take into account accents (eg. Protégé vs. protège);
• Can take into account structural context (HTML tags, links, table, paragraphe, etc.);
• Can take into account semantic context (eg. "apple" but not in the context "apple of discord")
• Filters documents that must contain (+) ou must not contain (-) particular concepts;
• Deployment on any system equipped with a Java virtual machine (Window, Linux, Unix).

WATCH PLATFORM, COLLECTION, PROCESSING, TARGETED DIFFUSION AND EXPLOITATION OF INFORMATION
Discover

Pertinence Information Network is a multilingual watch platform for the collection, processing, targeted diffusion and exploitation of information. Each document published on the platform is processed in real time and disseminated according to the domains of interest of the users. A private space is allocated to every user for the storage and organization of its documents in folders that can be shared with collaborators (e.g. bookmarks). This storage space is reserved for you to organize and archive your alerts and your documents.

Features :

• Multilingual : supports ten languages (West European and Asian languages, Arabic and Russian);
• Multiformat : supports various formats (HTML, PDF, DOC, RTF, TXT and PPT, XLS according to the platforms used);
• Handles XML documents (after automatic conversion);
• Management of a private workspace for every user to archive and organize personal work documents. This space is organized in embedded folders created by the user according to organization needs and optionally sharable with collaborators (e.g. bookmarks or text documents);
• Management of profiles and domains of interest;
• Management of privileges (security management);
• Publication system for content providers: possibility to create public or restricted-access diffusion channel, create privilege groups (e.g. to define subscribers groups), publish texts using simple cut-and-paste, entry, transmission of files or URLs;
• Management of diffusion channels according to the user's privileges (public channels, vs. restricted channels requiring a subscription or specific privileges). These channels can be organized hierarchically (e.g. as a thematic structure);
• Access to a customized news thread presenting all the new articles published in the sources selected by the user;
• Advanced alert system based on linguistic processing of the contents of published articles. Every user can define domains of interest by simply declaring concepts (keywords, proper nouns, etc.), set the update frequency and the composition of the alert messages, as well as several other options. The user then automatically receives highly targeted alerts, since all the referenced documents are relevant to a specific domain of interest. Each reference can also be accompanied by an automatically generated summary which can be customized by number of sentences, words or characters;
• Innovative feature for the exploitation of content, based on our advanced summarization technology. Through a single mouse click, the user can open a document from the alert hit list, highlight important sentences using a multi-level annotation system, highlight the concepts of his/her domain of interest, hide specific passages of the text, or derive a summary whose relative size can be freely defined – that is, exploit information efficiently;
• Support of the RSS (Rich Site Summary) format for the publication on the platforms documents from media sites.
• Meta search engine : includes Pertinence Meta Search which allows periodical web searches and search results filtering with pertinence operators;
• Information cartography : co-occurrence maps may be periodically generated to display simultaneous occurrence of particular concepts;
• Available in two versions: server and workstation. Java API and Web API services are available;
• Deployment on any system equipped with a Java virtual machine (Window, Linux, Unix).

Copyright © 2001-2006 Pertinence Mining