There are so many Business Intelligence (BI) solutions out there. How does Rosoka fit?

The Rosoka Multilingual Product Suite enables BI solutions to capture key pieces of information from their natural language holdings, such as newswire, emails, company reports, etc. regardless of the language. They can capture and correlate names like Warren Buffet in English, Chinese (沃伦·巴菲特), Russian (Уоррен Баффет), or Arabic (وارن بوفيه ); determine the relationships to him within the content and include salience and polarity of the context. Rosoka extracts key entities and their relationships.


How many entity types does Rosoka extract?

Rosoka is a highly customizable and flexible tool that can be configured to meet your analytics needs. We provide the vast majority of extraction needs, such as person, organization, place, facility, phone, date/time. Rosoka also comes with a robust set of relationships that connect these entities together, such as person-to-person, person-to-organization, person-to-place. If there are additional entities or relationships that are particular to your operational needs, Rosoka is easily customized. Every entity and relationship type that is provided or configured is given for every language. As a matter of fact, every language that Rosoka extracts is included, nothing additional needed with Rosoka.


What platforms does Rosoka run on?

Rosoka runs on any platform with a Java Virtual Machine (JVM), such as Windows, Linux, and Mac.


For Integrators, how would you rate the ease of integration and usability with clients that already have a solution in place, but want the added advantage of language identification and multilingual information?

Rosoka was designed to work well with others. Through the intuitive Java API or configurable web service interface, Rosoka easily fits into any pipeline.


What is the difference between processing in multiple languages, and being multilingual?

Being truly multilingual is more than just being able to process in several languages simultaneously. With Rosoka, if the language changes within a document, within a paragraph or within a sentence, it should not be skipped, dropped, or kicked-out from the process, but in many cases, where only one language is processed, it takes several runs and additional time re-processing to handle documents written in more than one language.

When a person is multilingual, they use their languages simultaneously within a phrase. Rosoka is multilingual enabling it to process, understand and follow linguistic rules in order to correctly extract relationships and entities. Like a great linguist, it’s all in there.

Rosoka reads, writes, speaks, understands and processes 230 languages, as if they were all part of one function. No extra time needed to re-process, no information dropped and meanings are extracted in their native language; nothing gets lost in translation.

Rosoka’s Language ID engine will determine the Native Language being used, then will process the meaning and the extraction within that Native Language. Reports and information presented in English won’t skip over any information as they will be simultaneously and seamlessly processed, without losing any of the relationship comparisons within the document. This will give you the most accurate salience, polarity, and Value Specific results for your entities and relationship reports.


Can I use Rosoka with my RDF triple store?

Yes, Rososka’s Entity Extraction output includes the option to output as RDF. Entity relationships are expressed as RDF triples and can be stored in open source (like Apache Jena TDB) or commercially available products (like MarkLogic) where they can be searched with SPARQL and displayed as relationship graphs.