Aspects from Appraisals!! A Label Propagation with Prior Induction Approach
Performance appraisal (PA) is an important Human Resources exercise conducted by most organizations. The text data generated during the PA process can be a source of valuable insights for management. As a new application area, analysis of a large PA datas
- PDF / 165,674 Bytes
- 9 Pages / 439.37 x 666.142 pts Page_size
- 14 Downloads / 196 Views
TCS Research, Tata Consultancy Services, Pune 411013, India {nitin.ramrakhiyani,sachin7.p,gk.palshikar,manoj.apte}@tcs.com 2 Department of CSE, Indian Institute of Technology Bombay, Mumbai 400076, India
Abstract. Performance appraisal (PA) is an important Human Resources exercise conducted by most organizations. The text data generated during the PA process can be a source of valuable insights for management. As a new application area, analysis of a large PA dataset (100K sentences) of supervisor feedback text is carried out. As the first contribution, the paper redefines the notion of an aspect in the feedback text. Aspects in PA text are like activities characterized by verb-noun pairs. These activities vary dynamically from employee to employee (e.g. conduct training, improve coding) and can be challenging to identify than the static properties of products like a camera (e.g. price, battery life). Another important contribution of the paper is a novel enhancement to the Label Propagation (LP) algorithm to identify aspects from PA text. It involves induction of a prior distribution for each node and iterative identification of new aspects starting from a seed set. Evaluation using a manually labelled set of 500 verb-noun pairs suggests an improvement over multiple baselines.
1
Introduction
Performance Appraisals (PA) are carried out in various organizations to measure growth and productivity of employees. Apart from gauging employee performance, the PA process is used for employee promotions, re-numerations and rewards. In this work, PA process of a large Information Technology services organization is considered. The organization has more than 300 K employees, leading to large amount of appraisal text (∼10 million sentences) getting generated every year. The organization’s PA methodology consists of three steps - (i) supervisor setting goals for an employee, (ii) employee recording his self-appraisal for each goal and (iii) supervisor recording his feedback. It is important to summarize such appraisal dialogue. For summarization in the product review domain, “aspect” based analysis is carried out. Aspects are certain informative noun phrases useful in summarization and sentiment analysis. A novel contribution of this work is the redefinition of an aspect in PA domain which is more an activity than a single c Springer International Publishing Switzerland 2016 E. M´ etais et al. (Eds.): NLDB 2016, LNCS 9612, pp. 301–309, 2016. DOI: 10.1007/978-3-319-41754-7 28
302
N. Ramrakhiyani et al.
noun phrase. As the second contribution, the Label Propagation algorithm [12] is supplemented to introduce an instance-wise prior which indicates an instance’s belongingness to the target labels. The approach starts with a seed set and then learns new aspects iteratively using weakly supervised label propagation.
2
Notion of an Aspect in the PA Domain
Based on a perspective to regard a performance appraisal as a kind of “person review”, the term “aspect” is borrowed from the domain of product review analysis. Observation
Data Loading...
