Development of medical records data science solutions. Tasks have included:• Designing and developing in Python a system which incorporates SVM classifiers to analyze charge capture audit physician CMS-1500 records in order to generate missing procedure code predictions.

Software development relevant to resume analysis and job candidate discovery. Developing the natural language functionality required by the company’s executive job search app, its social networking Throng app, and its Graph business contact discovery software. Tasks have included:• Designing and developing an information extraction system that extracts and normalizes from resumes contact information, educational background, employment history, etc.• Designing and developing a vector… Show more Software development relevant to resume analysis and job candidate discovery. Developing the natural language functionality required by the company’s executive job search app, its social networking Throng app, and its Graph business contact discovery software. Tasks have included:• Designing and developing an information extraction system that extracts and normalizes from resumes contact information, educational background, employment history, etc.• Designing and developing a vector space model classifier that assigns resumes to profession categories, and ranks them according to degree of expertise.• Designing and developing a vector space model classifier that interprets from a resume degree of management experience (e.g. executive, mid-level).• Designing and developing semantic matching functionality based upon WordNet resources for the company’s Throng app.• Designing and developing functionality which automatically summarizes and extracts salient terms from news articles for the company’s executive job search app. • Designing and developing functionality which returns a list of LinkedIn profiles similar to those of a user-provided set of sample profiles.• Designing and developing functionality which returns job function classifications (e.g. Marketing, Engineering) based on the job title words of a LinkedIn profile. Show less

Development of medical records NLP solutions. Tasks included:• Designing and developing in Java vector space model classifiers to assign partial ICD-10 PCS codes to medical documents containing descriptions of procedures.• Designing and developing in Java a rule-based Clinical Documentation Improvement process, which leverages the SNOMED ontology as well as expert knowledge to discover and alert for diagnostic information missing from a medical document.• Designing and developing… Show more Development of medical records NLP solutions. Tasks included:• Designing and developing in Java vector space model classifiers to assign partial ICD-10 PCS codes to medical documents containing descriptions of procedures.• Designing and developing in Java a rule-based Clinical Documentation Improvement process, which leverages the SNOMED ontology as well as expert knowledge to discover and alert for diagnostic information missing from a medical document.• Designing and developing, as a proof-of-concept, a semi-supervised process for highly accurate drug dosage and body metric numeric information extraction.• Designing and developing in Java a vector space model classification process of language identification for 23 Latin character languages. Show less

Data set compilation and optimization, accuracy testing, and software development relevant to the company’s web content classification and malware detection products. Tasks included:• Compiling and tuning, for 15 language versions, training sets for the company’s Naïve Bayes web page classifier.• Various research projects relevant to the classification of a web page’s language and content.• Designing and developing a classifier capable of detecting phishing websites.• Designing… Show more Data set compilation and optimization, accuracy testing, and software development relevant to the company’s web content classification and malware detection products. Tasks included:• Compiling and tuning, for 15 language versions, training sets for the company’s Naïve Bayes web page classifier.• Various research projects relevant to the classification of a web page’s language and content.• Designing and developing a classifier capable of detecting phishing websites.• Designing and developing vector space model behavioral classifiers capable of detecting malware executables based upon sandboxing logs.• Designing and developing a method capable of extracting unique n-grams indicative of malware from within the subdomain, domain, or path+filename of a URL.• Designing and developing a classifier capable of detecting adult-oriented websites based upon language use in the subdomain, or path+filename of a URL. Show less

Development of a variety of NLP software products. Tasks included: • Writing a regression testing application to guarantee the accuracy of Inxight's named entity information extraction software product. • Writing finite state transducer rules to normalize spelled and numeric date, currency, and percent surface forms to ISO representations. • Writing a web-based demo integrating all of Inxight’s linguistic software. • For Inxight’s MailFriday web-based email service, adapting… Show more Development of a variety of NLP software products. Tasks included: • Writing a regression testing application to guarantee the accuracy of Inxight's named entity information extraction software product. • Writing finite state transducer rules to normalize spelled and numeric date, currency, and percent surface forms to ISO representations. • Writing a web-based demo integrating all of Inxight’s linguistic software. • For Inxight’s MailFriday web-based email service, adapting existing and writing new software aimed at reducing the time required for an email user to process large amounts of mail. The new software, written in Java, included automatic categorization (implemented using KNN), meta-content labeling and indexing, automatic priority sorting, and contact information capture. • Assisting in porting a purchased Arabic tokenizer, analyzer, and stemmer from Perl scripts to finite state architecture. • Compiling the data and writing the finite state rules for Inxight’s first Arabic named entity extraction product, which extracts names of people, places, organizations, dates, currency expressions, etc. from Modern Standard Arabic text. Show less

• Co-production of video-oriented English as a Second Language multimedia titles. Tasks included: • Co-producing video for Cine-Bit clients.• Co-designing and co-producing a Cine-Bit title.• Supervising the localization of the Cine-Bit title into several European languages.

Computational lexicography and electronic publishing software development. Tasks included:• Designing and developing in Xbase a data entry system for lexicographers.• Compilation of monolingual and aligned bilingual corpora, and applying to them computer-assisted and fully-automated analyses to facilitate dictionary enhancement. • Project management and co-management of 10 electronic publications, including the Unabridged Larousse English-French/French-English CD-ROM and the… Show more Computational lexicography and electronic publishing software development. Tasks included:• Designing and developing in Xbase a data entry system for lexicographers.• Compilation of monolingual and aligned bilingual corpora, and applying to them computer-assisted and fully-automated analyses to facilitate dictionary enhancement. • Project management and co-management of 10 electronic publications, including the Unabridged Larousse English-French/French-English CD-ROM and the Larousse English-French/French-English Handheld. Show less

Computational linguistics and lexicography, and electronic publishing software development. Tasks included:• Designing and developing in Xbase a data entry system for lexicographers.• Data conversion and preparation for eight HarperCollins electronic publications.• Writing the Spanish generation module for an interlingua-based machine translation project.

Development of a database system to track cost savings projects.