Technology

At Language Computer, research is more than just our passion -- it's also how we provide unbeatable value to our customers.

Since our founding in 1995, we've worked hard to stay at the forefront of research in computational linguistics, information retrieval, information extraction, textual inference and knowledge management. It's only natural: we want to build the world's best products for annotating, searching, and understanding natural language.

Language Computer's products are supported by four different research and development teams: a text processing group which focuses on the annotation of English and foreign-language texts, an information extraction group devoted to the extraction of entities, attributes, relationships, and events mentioned in texts, a content personalization group which develops the core customization engines used in a number of LCC's products, and finally, a question answering group dedicated to creating a robust semantic search capability.

Text Processing

We believe that robust natural language understanding starts with high-performance text processing systems. LCC currently leverages a wide range of different text annotation components, including its CiceroLite named entity recognition system, PinPoint temporal and spatial normalization system, and within-document and cross-document coreference resolution systems.

Information Extraction

Unlike information retrieval systems, information extraction systems -- like LCC's CiceroCustom -- deliver value from texts by automatically identifying specific information about the entities, events, and relationships central to a particular domain of interest. Unlike most traditional approaches to information extraction, LCC has pioneered a new, open-domain, extraction framework which acquires information from unstructured texts with unsurpassed precision and recall without the need for hand-crafted rules or pre-specified extraction templates.

Social Insights

Our Social Insights Platform was created to understand the social nuances in posts on forums, blogs, and twitter. Our software is able to identify social actions such as, disagreement, affordance, influence, disrepectful behavior, and goal setting. The patterns of these actions within a conversation can indicate: 1) the social role of the speaker; 2) the social relationship between the speaker and audience; or 3) the level of motivation of the speaker.

Content Personalization

LCC's work in content personalization has enabled the development of a new generation of customizable information extraction systems which reduce the time and effort required to create new entity, attribute, relationship, or event extractors. Instead of limiting professional analysts and application builders to a select set of extractors, users of LCC's customizable information extraction tools can create new high-performance, domain-relevant extractors -- usually in 30 minutes or less.

Question Answering

Automatic question answering systems leverage a hybrid of information retrieval and natural language processing techniques in order to find answers to factual and complex questions submitted by users. Unlike the keyword search technologies used by most current commercial search providers, question answering systems semantically search documents in order to find the exact type type of information sought by the question.