Use of Laplacian Projection Technique for Summarizing Likert Scale Annotations

TL;DR

This paper introduces a novel graph-theoretic Laplacian projection method to effectively summarize Likert scale annotations from human annotators, aiming to improve the aggregation of subjective judgments.

Contribution

The paper presents a new Laplacian projection technique specifically designed for summarizing Likert scale annotations, along with an analysis of its properties using real datasets.

Findings

Effective summarization of Likert annotations demonstrated

Properties of the Laplacian projection method analyzed

Potential for improved human judgment aggregation

Abstract

Summarizing Likert scale ratings from human annotators is an important step for collecting human judgments. In this project we study a novel, graph theoretic method for this purpose. We also analyze a few interesting properties for this approach using real annotation datasets.