Text / Document Classification

We would like to hire a programmer / or a team of programmers to develop for us a text/document classification tool that will be integrated into a system we are developing. The tool should be capable of classifying unstructured text (natural language text) into various categories. The user should be able to choose a classification method from an optional list of techniques provided by the tool (for example, logistic regression classifier, naïve Bayesian classifier or support vector machines). Examples of text/document classification include sentiment analysis suggesting categories for indexing a document created in a content management system (e.g. classifying news articles into categories like sports, politics, entertainment etc.), which we will discuss with the contracted individual/team. We welcome using any programming language as long as it is adopted into a Flex environment. For example the code could be written in Java and the Java classes can be utilized within the Flex environment. The front end will be a simple Flex/Flash user interface (see the attachment as an example). First the application will be trained using some sample data. Users will then insert text into the interface, and this text will be classified by the application. Interested parties should provide us with the following: - Background information highlighting skill and experience in NLP / Computational Linguistics technologies - A summary list of systems/projects of similar nature they have worked - 2-3 references of prior clients we can contact and verify information - Statement of interest Statement of Interest should be 2-4 pages in length, and shall include sufficient details about the following: - Your understanding of the task and the purpose of the tool you will develop. - Your approach to tool development: Explain the approach you propose to use for developing this tool. Provide a justification for what methods the tool is going to support and when and why a user may choose one over another. - Performance: The tool may be used to screen and classify several thousands of documents at once, so you will also need to discuss performance issues associated with each method. - Timeline: How long would it take to you to complete; and a breakdown of milestones including a plan for testing code. We are aware that there are so many pieces of code and libraries available (such as GATE, SVMfu etc.). So if you are planning to utilize an existing piece of code or libraries, you will need to make sure it is free to use for commercial purposes, and let us know that you are planning to use it as well. In fact, we suggest following one of the existing scientific methodologies (such as [login to view URL]), make use of existing code as much as possible, and re-write necessary portions to make Flex integration possible. Feel free to use visual aids to communicate your ideas to us. It would help if you can come up with some dummy screenshots, flowcharts and graphics of solution architecture. We have a product plan that entails more tools related to this area, so we are looking for programmers who we can build with an extended working relationship. Requirements - At least a past programming project done in Natural Language Processing - Some level of education in Natural Language Processing The end product will include: - Source code and code documentation - Simple Flex interface - Product demo - Test report Budget will be commensurate with experience and clarity of statement.