Order Out of Chaos: Hybrid Model for Administrative Document Sorting and Annotation classification

Condo Clear is a company that is in the business of simplifying the Strata corporation documents, helping the clients to make thoughtful decisions while purchasing real estate. What makes this task of summarizing information so cumbersome is the number of documents associated with each corporation and the distinct structure. The current process involves:
(a) manually sorting the set of documents of a given condo corporation in a standard order
(b) annotating the relevant information for the creation of a summary for the client
(c) assigning class labels to the annotated text
The present scenario limits the ability of the company to meet the demand of the ever-growing real estate market. The goal of the project is to replace manual work with an interactive system that can perform these tasks with human supervision. The condo corporation documents are based on templates with minor variations specified by the Strata law, customized for the individual corporation, and free text documents, e.g., minutes of meetings. This project will integrate deep language models for free text with traditional NLP and rule-based pattern-matching techniques to address (a) and (c).

Faculty Supervisor:

Evangelos Milios

Student:

Partner:

Condo Clear Services Inc.

Discipline:

Computer science

Sector:

Real estate and rental and leasing

University:

Dalhousie University

Program: