80 likes | 178 Views
Notes on the Data Quality Problem in Location-Based Services. Artem Katasonov WIM network workshop, Uppsala September 2004. Goals of this presentation. To present, in very general terms, those of the problems I deal in my doctoral dissertation, which are of interest for WIM participants.
E N D
Notes on the Data Quality Problem in Location-Based Services Artem Katasonov WIM network workshop, Uppsala September 2004
Goals of this presentation • To present, in very general terms, those of the problems I deal in my doctoral dissertation, which are of interest for WIM participants. • To introduce the audience into an issue, to which they probably did not ever pay attention before. To sketch some research directions.
Wireless Information Management requirements acquisition communication quality assurance What? Data management(http://www.acq.osd.mil/io/se/cm&dm/) is the process of applying policies, systems and procedures for: • identification and control of data requirements; • for the timely and economical acquisition of such data; • for assuring the adequacy of data for its intended use; • for the distribution or communication of the data to the point of use; • and for use analysis.
Location-based services An LBS is a Ubiquitous Decision Support System (UbiDSS) High dependability requirement Possible negative impact Dependability requirement Why? User Value? - Mobile services: information ”anywhere” and ”anytime” - LBS: information relevant “right here” and “right now” Why to use? - To make immediate decisions, e.g. about where to go
Why? (2) In a hospital, a nurse is writing down a mail from the dictation of the man, who is completely covered by plaster, with arms and legs broken: Dear editor, I would like to inform you that on the page 14 of your manual “How to handle a helicopter” I discovered a minor misprint... Providing information at right place and time is a great business opportunity. However, With this great opportunity comes great responsibility.
inaccurate non-existing missing redundant Why? Practice • LBSs developed in small companies, e.g. by mobile operators • Data repurposed, e.g. from phone books Case - we helped one of Finnish operators in performing content quality evaluation of their LBS “find the two nearest facilities of a type”, data is from a yellow-pages publisher: • Evaluated as statistical probability of getting correct answer, quality is well below a level that could be considered sufficient • Only a small part is because of ”bad work” of the content provider • Most is direct result of using data repurposed from a phone book: • Not collected proactively – many omissions • Updated once a year • Some categories are too broad and some are or too narrow • Facilities rarely counted into several categories
Some Research Directions • How to measure data quality in LBS (and at a reasonable cost)? • How to achieve sufficient quality of data (and at a reasonable cost)? • How to deal with repurposed data? • How to balance utility with quality achievable? Some conceptual and business questions: • How data quality influences adoption of LBSs? • How much quality may cost? • Who in the business chain/network should be responsible for quality?