"This could help us understand what subsequent actions the user is likely to take for example, and of course understand where negative feelings arise and allow us to reduce them."
This is an area of research which has been left behind and is still in its infancy. Explicit and implicit feedback have been used and researched however they have very real limitations which affective computing may be able to address.
There's a paper that was presented at SIGIR 2008 called "Affective Feedback: An Investigation into the Role of Emotions in the Information Seeking Process" by Ioannis Arapakis, Joemon M Jose and Philip D Gray. Here is a summary of the main points:
The current techniques which are used for relevance feedback (namely Explicit and implicit) can determine content relevance according to the cognitive and situational levels of interaction between the user and the retrieval system. The problem with this is that they don't take into consideration user intentions, motivations, feelings and so on which can affect their information retrieval behaviour.
The emotional responses observed vary widely from user to user and from situation to situation. This means that there needs to be a method which is capable of dealing with this.
Implicit and explicit feedback can determine whether a document is relevant or irrelevant. For Explicit feedback there's a trade-off between getting the documents that the system sees as important and those that the user are genuinely interested in. "Eventually, as the task complexity increases the cognitive resources of the users stretch even thinner, turning the process of relevance assessment into a non-trivial task."
Additionally implicit feedback whilst collecting information about the user search behaviour suffers from reliability issues. What can be observed does not match the user intention. Belkin and Kelly showed that implicit feedback is "unreliable, difficult to measure and interpret".
Kuhlthau found that there were 3 dimensions: affective, cognitive, and physical. The authors measure the physical using a range of biometric measurements (GSR, skin temperature, etc.). They used a facial expression recognition system and applied hidden recording (because they wanted to be as invisible as possible).
They used the Indri opens source search engine from the Lemur project because it can parse TREC newswire and web collections and return results in the TREC standard format and it's also very reliable.
The results were that happiness and irritation were the most intense emotions - followed by sadness, pleasure and surprise.
"Task difficulty and complexity have a significant effect on the distribution of emotions across the three tasks. As the former increase, so do the negative emotions intensify and progressively overcast the positive ones. We hypothesize that this progression is the result of an underlying analogy between the aforementioned search factors and emotional valence, and, furthermore, that it is indicative of the role of affective information as a feedback measure, on a cognitive, affective and interactional level".
They do believe however that low-frequency scores may be more important compared to those with higher scores. They also found that "affective feedback should be treated differently as the task difficulty increases".
This is encouraging research because it does begin to address the issues in explicit and implicit relevance feedback. It also stays in line with Nicholas Belkin's request for more affective computing research!
Why should you care?
well if you imagine this system widely implemented across many search engines you would need to take into consideration psychology and cognitives when designing, structuring and optimizing your website. Knowing where users are going to go next in a search for documents affects your keyword research a great deal. This does throw a whole load of new variables into that task.
Subscribe by email. Enter your email address below: 
