I don’t want the search engine to think – I want the search engine to do free association. Steve Robertson.
Google: Prediction and assistance of the user with a much more conversational element
RBY: Spoken language to text queries
SR: The web is not going to last forever
SR: The battle between web search engines and web content providers – the future is not comfortable in that respect. Engineering of content by providers is problematic to search engines.
Belkin (2008) outlined some of the challenges within IIR
Jarvelin (2011) also argued the need to understand Info. Sys. Through the development of formal models and testable theories to describe the interaction b/w users and systems.
It is a major research challenge because of all the complexities involved with users, their interactions with information and the systems that they employ.
A firm produces output (such as goods or services) A firm requires inputs (such as capital and labor) A firm utilizes some form of technology to then transform the inputs into outputs.
Let’s think about what this formula is saying. The amount of gain or overall performance is a combination of A times Q. Imagine if all these were relevant, and you got 1 point per relevant document, than you would essentially obtain A times Q gain. However the inclusion of Alpha, generally means you get less as there is a trade-off between A and Q.
What strategy should a user eYes, the model is abstracted and general. ….. Search has many more inputs and outputs. Lots more variables, but we have abstracted away these details. We have simplified the search process to two core variables that affect the output. But this doesn’t mean the model doesn’t have any explanatory power Representative, but not necessarily wholly realistic
employ to achieve their goal?
And if we map a cost function to the interactions then we can ask, “what is the most cost-efficient way for a user to interact with an IR system?” What strategy should a user employ to achieve their goal?
A user can choose from a range of information seeking strategies The user’s time is an economic quantity i.e. cost The user pursues a particular strategy until the cost incurred exceeds the utility received, At this point the user may choose another strategy Or they stop
So for example on the structured interface, the user would have to: Go to the search box, type in the first query term, go to the next query box, type in a term, etc, then click search button and wait for the response. On the standard interface, the user would have to: Go to the search box, type in the three query terms, hit enter (activating the search) and then wait for the response, On the suggestion interface, the first query is the same, but subsequent queries, they could simply click on a query suggestion.
there is a measure of the output of the system, as well. As opposed to measures it. These models decouple the cost and benefit and parameterize them on the interactions. This means we can tease out how the interactions functionally related to each other.
Though work by Turpin and Hersh or Cantor and Smith have shown that on a degraded system, users adapted to degraded systems by posing less queries. This seems to fit with the reduction in the efficiency of the system (i.e. the k parameter)
1. Leif Azzopardi
Modeling Interaction with Economic
Models of Search
2. Interactive Information Retrieval needs formal models to:
• describe, predict and explain information behaviors
• provide a basis on which to reason about interaction,
• understand the relationships between interaction,
performance and cost,
• help guide the design, development and research of
information systems, and
• derive laws and principles of interaction
• e.g. Law of Least Effort In Finding
Major Research Challenge
3. INITIAL ECONOMIC MODEL OF SEARCH
4. All models are wrong
but some are useful
George E.P. Box
5. Production Theory
Applied to Searching
6. Gain Function for the Search Process
Let the gain the user receives through their
Q is the number of queries, and
A is the number of documents examined per query.
α is the relative efficiency of querying to assessing
k is the efficiency of the technology/user to extract/
identify relevant information returned
7. Gain Curve
Each point on the curve represents a combination
of interactions that will yield the same gain.
8. Cost Function for the Search Process
The total cost can be calculated by:
– cq is the cost of a query
– ca is the cost of a assessing a document
– A.Q is the total number of documents assessed
10. Few Queries,
What strategies can the user employ
when interacting with the search system to achieve their end goal
What course of action minimizes
the cost of interaction?
11. Cost Curve
The total cost is minimized when A = 10, which
corresponds to Q = 18. Any other combination will
result in a higher total cost.
12. How does behavior change
when query cost increases?
13. Query Cost vs Interaction
0 2 4 6 8 10
Azzopardi, Kelly & Brennan (2013)
as the cost of querying increases,
more documents will be examined per query,
and less queries will be issued
14. Testing the Query Cost Hypothesis
Azzopardi, Kelly & Brennan (2013)
Q = 19
A = 5
Q = 35
A = 1.6
A = 2.5
Structured vs Standard and Suggestion : YES
Standard vs Suggestion: NO
Model does not account for the time spent on the search
result page nor the interaction with snippets.
• Assumes users are rational
• Assume interaction is fixed
• Model of interface too simplified, the search
process is more than just querying and
– There are lots of other costs involved when
– There are lots of other interactions that can be
16. A NEW
ECONOMIC MODEL OF SEARCH
17. Modeling Other Costs
Cost to enter a query (cq)
Cost to load search page per query
Cost to examine each snippet
Cost to view a document
Cost of return back to search page
Cost to assess the document (ca)
Cost to view next page
18. Modeling Other Costs
• Let’s also include the:
– cost of viewing pages (cv) and
– cost of examining snippets (cs)
in the cost model, such that:
• Let’s assume that the number of page views is
equal to some constant v
– Typically this would be v=1
– But could be the average number of pages
examined i.e. v=1.1
• Let’s further assume that A = S.pa
– Where pa is the probability of assessing a
document given a snippet.
20. Reducing the Cost Function
• Given these assumptions, the cost function
can be simplified down to the following:
21. New Gain Function
• Previously, Q and A were linked via α and 1-α,
• Here we decouple this relationships
– which enables us to estimate the parameters
– and so becomes more intuitive
22. Optimization Problem
• Given our model, we wish to minimize the
cost c(Q,A), subject to the constraint that
g(Q,A) = g
• To do this we used a Lagrangian multiplier
23. Optimal Interaction
The optimal number of assessments per query:
The optimal number of queries:
24. How does querying behavior change?
• So we can say more precisely that:
– If g increases then Q will go up
– If k increases then Q will go down
– If β increases, then Q will go down
– If α increases, then Q will go up
25. Some Cost Hypotheses
• Document Cost Hypothesis: as the cost of
document increases, Q increase, A decreases.
• Snippet Cost Hypothesis: as the cost of
examining snippets increases, A decreases,
while Q increases.
26. Performance Hypotheses
• Beta-Performance Hypothesis:
as β increases, A will increase,
while Q will decrease.
27. Assessment Probability Hypothesis
• Assessment Probability Hypothesis:
as the probability of assessment increases,
A increases, while Q decreases.
28. ACTUAL VERSUS OBSERVED
29. Analysis of Empirical Data
• Re-examined the experimental data from
Azzopardi, Kelly & Brennan (2013).
• Where we considered the different
interactions over topics for each condition
• And tested seven of the hypotheses that we
30. Beta Interaction Hypothesis
• Hypothesis states as β increases,
Q will decrease and A will increase.
• Observations tend to match theory
31. Assessment Probability Hypothesis
• Hypothesis states that as pa increases, Q
decreases, A increases
• Observations match theory
• Similar finding for snippet cost hypothesis
32. Document Cost Hypothesis
• Hypothesis suggests that as Cd increases, Q
should increase, while A should decrease
• But clearly this is not the case.
33. An explanation
• Cs / pa dominates Ca, so when considered
together, the result matched our expectation
• i.e. Cs / pa is bigger than Ca, and thus has a
greater influence on the results.
34. An explanation
• As increases, then Q should
increase, while A should decrease.
• Considering all three variables we see that this
tends holds in practice.
35. Summary of Empirical Findings
• The new model generally fits the empirical
– When there were deviations, we could explain
these through other variables having a greater
influence on the interaction
• β tends to dominate interaction
– Ie. Low β, leads to fewer documents being
assessed per query
– cs and pa also play a major role in shaping
• This new models provides a better description
of the search process
• By framing IR Tasks as economic optimization
problems we can derive testable hypotheses!
• These models provide the functional
relationships between interaction,
performance, and cost and how it affects
37. Open Questions
• How well does the theory match up to
• How do these economic models relate to
Information Foraging Theory or the
Interactive Probability Ranking Principle?
• What happens when users are not rational?
• What other insights can we obtain when
applying this approach to other IR task?
38. In theory, theory and practice are
the same. In practice, they are not.