This chapter focuses on the science of human perception of information quality and describes a subset of Information Quality (IQ) dimensions, which are termed Subjective Information Quality (SIQ). These dimensions typically require a user’s opinion and do not have a clear mathematical technique for finding their value. Note that most dimensions can be measured through multiple techniques, but the SIQ ones are most useful when the user’s experience, opinion, or performance is accounted for. This chapter explores SIQ while considering information obtained from multiple sources, which is a common occurrence when employing visualizations to perform business or intelligence analytics. Thus, the issues addressed here are the assessment of subjective perception of quality of data shown through visual means and principles on how to estimate the subjective quality of combined information sources.
TopIntroduction
Modern analytics tools employed in business, intelligence, and even science, rely increasingly on rapid exploration of multiple information spaces and on a combination of various information dimensions and sources performed by an analyst. Integrating information quality involves providing users with a unified view of this data quality. This process becomes significant in a variety of situations both commercial and scientific, particularly with the widespread use of information visualization to allow end-users prompt access to large amounts of data. Integrating information appears with increasing frequency as the volume and the need to share existing information grows.
Information quality (IQ) is not a simple scalar measure, but can be defined on multiple dimensions, with each dimension yielding different meanings to different information consumers and processes. Each dimension can be measured and assessed differently. Information Quality assessment implies providing a value for each dimension about how much of the dimension or quality feature is achieved in order to enable adequate understanding and management.
This chapter focuses on a subset of IQ dimensions, which we term subjective information quality (SIQ). These dimensions typically require a user’s opinion and do not have a clear mathematical technique for finding their value. Note that most dimensions can be measured through multiple techniques, but the SIQ ones are most useful when the user’s experience, opinion, or performance, are accounted for. The objective dimensions are the dimensions that can be largely and commonly assessed via mathematical or functional forms. It is possible for the same dimension to be measured by both subjective and objective means, depending on the context. For example, accuracy can be objectively computed when one checks the contents of a warehouse against an electronic inventory record. However, accuracy can also be subjectively employed, as for example, the estimation of the amount of shade in a specific area. Table 1 shows all IQ dimensions and the relative importance or usefulness of objective metrics versus subjective ones (Pipino, Lee, & Wang, 2002).
Table 1.
IQ dimensions and the relative usefulness of objective metrics versus subjective ones
| Dimension | Objective Assessment (width of gray area proportional to relevance of objective metrics) | Subjective Assessment (width of white area shows relevance for subjective) |
| Believability | Apply a formula (integration only) | User’s opinion and experience determines whether they trust the data |
| Ease of Manipulation | Time to perform a computation | User’s experience or performance with the data |
| Interpretability | Whether some computation is successful | User can understand the data correctly |
| Relevancy | Can produce a valid result | Helps the user in their task |
| Reputation | Apply a formula (integration only) | The user can judge or assume the accuracy based on the result of the objective assessment |
| Value-Added | Can increase the value of data | The user can judge or assess the value added to the data |
| Objectivity | Formulas applied | User opinion |
| Timeliness | Can reflect how up-to-date the data is with respect to the task | User judges based on previous experience |
| Understandability | Can provide clear and simple data | User can understand the data easily |
| Concise Representation | The shortest representation is known | User judges based on previous experience |
| Appropriate Amount of Data | The needed amount is known | User expertise is required |
| Security | Against a standard metric | Users experience or performance with the data |
| Accessibility | Against a standard metric | Based on user’s experiences |
| Consistent Representation | Count different representations | User’s opinion |
| Accuracy | Formula based on known, exact value | Expert estimation when exact value not available |
| Completeness | Count missing values in structured sources | User’s opinion for unstructured text |