Enabling knowledge accessibility for a customer support unit with an information retrieval portal

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Master's thesis

Date

2020-10-20

Department

Major/Subject

Data Science

Mcode

SCI3095

Degree programme

Master's Programme in ICT Innovation

Language

en

Pages

82+12

Series

Abstract

Customer support units generate massive amounts of free-text data. The retrieval of this information can be of great interest for companies to provide better customer support services. The customer support units in manufacturing companies have particular requirements in this aspect, sometimes requiring complex technical information from past cases. However, the information tools available for in this customer support units sometimes can prove to be inefficient and complicated from the workers' point of view. This thesis explores how an information retrieval portal, built as an Minimum-Valuable-Product, provides relevant information accessibility to a customer support unit. A full-stack solution was developed after analysing the workers' information needs, knowledge tools and state-of-the-art technologies in information retrieval. The technology used stores in a No-SQL information retrieval storage system which provides a new data scheme, and a boolean scoring function is used to retrieve the documents. The solution is evaluated from two points of view: from an information retrieval point of view by measuring the precision of the first ten results retrieved, and from a user experience perspective, to measure the learnability and user satisfaction of the new tool in comparison with the old one . As the main results of this thesis, we show that the boolean query strategy was suitable for the information needs of the users. The precision of the retrieval system achieved good results. The design of the portal provided a simple and easy user interface for users as it is reflected in the learnability results. In general, the solution was well received by the users as the commentaries provided show. Nevertheless, further improvements can be made by addressing some problems design flaws that the users flagged.

Description

Supervisor

Kurimo, Mikko

Thesis advisor

Saynevirta, Simo

Keywords

information, retrieval, search, engine, elasticsearch, interface

Other note

Citation