Scribd
Upload
57 views1 page

Task Analysis Compared

This document discusses using a Bayesian machine learning approach to task analysis. It presents a case study analyzing conversations between customers and agents at a call center. The conversations were recorded and used as input for a Bayesian tool to analyze the task. The tool can output subtask frequencies and durations, timelines, process diagrams, and classifications to understand the agent's troubleshooting work. A model is shown where customer and agent data streams like audio, video, and computer interactions are synchronized and fed into the Bayesian tool for analysis.

Uploaded by

Srinivas
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
57 views1 page

Task Analysis Compared

This document discusses using a Bayesian machine learning approach to task analysis. It presents a case study analyzing conversations between customers and agents at a call center. The conversations were recorded and used as input for a Bayesian tool to analyze the task. The tool can output subtask frequencies and durations, timelines, process diagrams, and classifications to understand the agent's troubleshooting work. A model is shown where customer and agent data streams like audio, video, and computer interactions are synchronized and fed into the Bayesian tool for analysis.

Uploaded by

Srinivas
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

A Bayesian Based Machine Learning Application to Task Analysis

MAIN FOCUS environment (Zsambok, 1997; Klein, 1998; Todd &


Gigerenzer, 2001; Hutton et al, 2003). The following
The Bayesian framework provides a potentially more discussion presents a brief case study illustrating the
applicable method of task analysis compared to com- application of a Bayesian method to task analysis.
peting approaches such as neural networks, natural This particular study here focuses on describing what
language processing methods, or linguistic models. Two takes place in a call center, when the customer calls to
Bayesian methods are often proposed: naïve Bayes and report various problems and the knowledge agent helps
fuzzy Bayes. Over the decades, studies such as those troubleshoot remotely. In this example, the conversa-
of Bookstein, (1985), Evans and Karwowski (1987), tion between agent and customer was recorded and
Lehto and Sorock (1996), Chatterjee (1998), Yamamoto manipulated to form a knowledge database as input to
and Sagisaka (1999), Zhu and Lehto (1999), Qiu and the Bayesian based machine learning tool.
Agogino (2001), Hatakeyama et al. (2003), Zhou and
Huang (2003), Leman and Lehto (2003), Wellman et Model Development
al. (2004), and Bolstad (2004) have shown that statisti-
cal machine learning within the framework of fuzzy Figure 1 illustrates important elements of the dialog
Bayes can be more efficient when the assumptions of between a call center knowledge agent and customer.
independence are violated. McCarthy (2002) found that The arrows indicate data flow. The dialog between the
fuzzy Bayes gave the highest success rate for print defect customer and the knowledge agent can be recorded
classification compared to ID3, C4.5, and individual using several methods. For example, if the customer
keyword comparison algorithms. Noorinaeini and Lehto uses e-mail, these conversations are directly available
(2007) compare the accuracy of three Singular Value in written form. The knowledge agent’s troubleshoot-
Decomposition (SVD) based Bayesian/Regression ing processes similarly could be recorded in video
models and conclude that all three models are capable streams, data screens, time-stamp streams of keystrokes,
of learning from human experts to accurately categorize mouse-clicks, data streamed to the agent’s monitor,
cause-of-injury codes from injury narrative. or various forms of data entry used by agents. These
Case studies have contributed to both theoretical and data streams can be synchronized with a time-stamp
empirical research in the naturalistic decision making as input for the Bayesian based machine learning tool.

Figure 1, Model of Bayesian based machine learning tool for task analysis1

Customer’s calls/e-mails Agent’s troubleshooting via phone/emails

Video streams, data screens, time-stamp streams of


keystrokes, mouse-clicks, data streamed to the
Audio streams, e-mail (if any) agent’s monitor, or other forms of data entry used
by agents

Synchronized various forms of time stamped video/audio/data streams


input
Bayesian based machine learning tool for task analysis
output
Decomposed subtask frequencies and duration; tool frequency and duration, timeline analyses; operational
sequence diagrams; hierarchical problem classifications; agent's solution classifications



You might also like