Browse Prior Art Database

Cognitive feedback analysis to proactively trigger automatic system modifications and user actions

IP.com Disclosure Number: IPCOM000247630D
Publication Date: 2016-Sep-21
Document File: 9 page(s) / 73K

Publishing Venue

The IP.com Prior Art Database

Abstract

This article describes an approach to automatically analyse sources where users talk about / report issues about a system / service by combining keywords and tone analysis, in order to take measures that either solve the potential issue before it is officially reported or that enable an effective analysis as soon as official support requests are received. Using this approach, it is possible to automatically identify a general problem that exists with a system / service even before users complain via the established support process, react faster by taking measures that either solve the issue or allow for an effective analysis as soon as reports from users arrive via the support process, and to automatically modify the system or its settings based on the findings.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 17% of the total text.

Page 01 of 9

Cognitive feedback analysis to proactively trigger automatic system modifications and user actions

Every system that provides functionality to users can encounter issues. These could for example be:

Performance issues: The response / throughput / etc. of the system isn't good enough.

Functionality issues: A specific functionality doesn't work at all or isn't working as expected.

Availability issues: The system or subsystem cannot be reached.

The more users are affected, the more critical it is to be informed as early as possible that an issue might exist in order to be able to start the problem analysis and / or take corrective measures. This is true for products running on-premises, but even more so for systems hosted for a customer. In the era of Cloud services, problems with a company's service will not only affect SLAs and the relationship with the customer, but will also affect the reputation of the company as a reliable Cloud service provider.

Every company providing products / services has a process in place with which users can report problems. Such a problem report will then successively be analyzed by L1, L2, L3 and potentially development. Each problem report describes a single issue by a single user / customer. If a general problem exists for a specific product / service, then this will likely affect many users and will therefore lead to a series of problem reports.

The challenges with this setup are:

There is often a delay between the user discovering a problem and submitting a formal problem report. At the same time, the earlier it is known that a problem might exist the better, as it might then be possible to solve the issue before even more users are affected by it. So by having to wait until problem reports are opened, support loses valuable time in which it would be possible to already initiate measures that will either solve the issue (e.g. by disabling a setting that had a negative impact on performance) before it affects other users, or to allow to analyze the issue more effectively (e.g. by enabling traces).

All the problems that are reported are analyzed as separate issues. If multiple users have the same problem, i.e. if there is a fundamental issue with the installation / the version of the product, then this doesn't become apparent right away.

The idea is to automatically analyze sources where users talk about / report issues about a system / service by combining keywords

keywords and tone analysis

                             tone analysis, and then to take measures that either solve the potential issue before it is officially reported or that enable an effective analysis as soon as official support requests are received.

Sources can include for example forum discussions, twitter posts, emails, blogs, etc.

These sources are automatically analyzed and checked for certain keywords, for example to identify a certain product or a class of problems. In addition, a cognitive capability like a tone analyser

          tone analyser is used to de...