Tefko Saracevic 1

RELEVANCE?RELEVANCE?in information science

Tefko Saracevic, Ph.D.tefkos@rutgers.edu

Tefko Saracevic

Two worlds in information science IR systems offer as

answers their version of what may be relevant by ever improving algorithms

People go their way & assess relevance by their problem-at hand,

context & criteria

The two worlds interact

2

Covered here: human world of relevanceNOT covered: how IR deals with relevance

Tefko Saracevic

Relevance interaction

3

URLs, references, and inspirations are in Notes

Tefko Saracevic

“Our work is to understand a person's real-time goal and match it with relevant information.”

4

“... relevant information.”

“... relevant ...”

Tefko Saracevic

Definitions

5

Merriam-Webster Dictionary Online

“1a: relation to the matter at hand b: practical and especially social applicability : pertinence <giving relevance to college courses> 2: the ability (as of an information retrieval system) to retrieve material that satisfies the needs of the user.”

Tefko Saracevic

Relevance – by any other name...

Many names e.g.“pertinent; useful; applicable;

significant; germane; material; bearing; proper; related; important; fitting; suited; apropos; ... “ & nowadays even “truthful” ...

Connotations may differ but the concept is still relevance

"A rose by any other name would smell as sweet“ Shakespeare, Romeo and Juliet

6

Tefko Saracevic

What is “matter at hand”? Context in relation to which

a problemproblem is addressed an information needinformation need is expressed

a questionquestion is askedan interactioninteraction is conducted

There is no such thing as considering relevance without a context

7

Axiom: One cannot not have a context in information interaction.

Tefko Saracevic

contextcontext – information seeking – intent

“Context – circumstance, setting:The set of facts or circumstances that surround a

situation or event; “the historic context” “ Wordnet

However, in information science & computer science as well:

8

from Latin: contextus  "a joining together” contexere  "to weave together”

“There is no term more often used, less often defined and, when defined, defined so variously, as context. Context has the potential to be virtually anything that is not defined as the phenomenon of interest.”

Dervin, 1997

Tefko Saracevic

context – information seeking information seeking – intent

Process in which humans purposefully engage in order to change their state of their state of knowledgeknowledge (Marchionini, 1995)

A conscious effort to acquire information in response to a need or gap in your knowledgea need or gap in your knowledge (Case, 2007)

...fitting information in with what one already fitting information in with what one already knows knows and extending this knowledge to create new perspectives (Kuhlthau, 2004)

9

Tefko Saracevic

Information seeking concentrations Purposeful process [all cognitive] to:

change state of knowledge respond to an information need or gap fit information in with what one already knows

To seek information people seek to change the state To seek information people seek to change the state of their knowledgeof their knowledge

Critique: Broader social, cultural, environmental … factors not included

10

Tefko Saracevic

context – information seeking – intentintent

Many information seeking studies involved TASK as context & accomplishment of task as intentintent

Distinguished as to simple, difficult, complex ...

But: there is more to a task then task itself time-line: stages of task; changes over time

11

Tefko Saracevic

Two large questions

Why did relevance become a central notion of information science?

What did we learn about relevance through research in information science?

12

Tefko Saracevic

WHY RELEVANCE?A bit of history

13

Tefko Saracevic

It all started with

Vannevar Bush: Article “As we may think” 1945 Defined the problem as “... the massive task of

making more accessible of a bewildering store of knowledge.” problem still with us & growing

Suggested a solution, a machine: “Memex ... association of ideas ... duplicate mental processes artificially.”

Technological fix to problem

14

1890-1974

Tefko Saracevic

Information Retrieval (IR) – definition

15

Term “information retrieval” coined & defined by Calvin Mooers, 1951

“ IR: ... intellectual aspects of description of information, ... and its specification for search ... and systems, technique, or machines...[to provide information] useful to user”

1919-1994

Tefko Saracevic

Technological determinant

In IR emphasis was not only on organization but even more on searching technology was suitable for searching

in the beginning information organization was done by people & searching by machines

nowadays information organization mostly by machines (sometimes by humans as well) & searching almost exclusively by machines

16

Tefko Saracevic

Some of the pioneers

at IBM pioneered many IR computer applications first to describe searching

using Venn diagrams

at Documentation Inc. pioneered coordinate indexing first to describe searching

as Boolean algebra

17

Mortimer Taube1910-1965Hans Peter Luhn 1896-1964

Tefko Saracevic

Searching & relevance

Searching became a key component of information retrieval extensive theoretical &

practical concern with searching

technology uniquely suitable for searching

And searching is about retrieval of relevant answers

18

Thus RELEVANCE emerged as a key notion

Tefko Saracevic

Why relevance?

Aboutness A fundamental notion

related to organization of information

Relates to subject & in a broader sense to epistemology

Relevance A fundamental notion

related to searching for information

Relates to problem-at-hand and context & in a broader sense to pragmatism

19

Relevance emerged as a central notion in information science because of practical & theoretical concerns with searching

Tefko Saracevic

WHAT HAVE WE LEARNED ABOUT RELEVANCE?

Relevance research

20

Tefko Saracevic

Claims & counterclaims in IR

Historically from the outset: “My system is better than your system!”

Well, which one is it? Lets test it. But: what criterion to use? what measures based on the criterion?

Things got settled by the end of 1950’s and remain mostly the same to this day

21

Tefko Saracevic

Relevance & IR testing

In 1955 Allen Kent & James W. Perry were first to propose two measures for test of IR systems: “relevance” later renamed

“precision” & “recall” A scientific & engineering

approach to testing22

Allen Kent1921 -

James W. Perry1907-1971

Tefko Saracevic

Relevance as criterion for measures

Precision Probability that what is

retrieved is relevant conversely: how much junk is

retrieved?

Recall Probability that what is

relevant in a file is retrieved conversely: how much relevant

stuff is missed?

23

Probability of agreement between what the system retrieved/not retrieved as relevant (systems relevance) & what the user assessed as relevant (user relevance)where user relevance is the gold standard for comparison

Tefko Saracevic

First test – law of unintended consequences Mid 1950’s test of two

competing systems: subject headings by Armed

Services Tech Inf Agency uniterms (keywords) by

Documentation Inc. 15,000 documents

indexed by each group, 98 questions searched

but relevance judged by each group separately

First group: 2,200 relevant Second: 1,998 relevant

but low agreement Then peace talks

but even after agreement came to 30.9%

Test collapsed on relevance disagreements

24

Results:

Learned: Never, ever use more than a single judge per query.

Since then to this day IR tests don’t

Tefko Saracevic

Cranfield tests 1957-1967

Funded by NSF Controlled testing:

different indexing languages, same documents, same relevance judgment

Used traditional IR model – non-interactive

Many results, some surprising e.g. simple keywords “high

ranks on many counts”

Developed Cranfield methodology for testing

Still in use today incl. in

25

Cyril Cleverdon 1914-1997

TREC started in 1992, still strong in 2014

Tefko Saracevic

Tradeoff in recall vs. precision

Example from TREC:

26

Generally, there is a tradeoff: recall can be increased by

retrieving more but precision decreases

precision can be increased by being more specific but recall decreases

Some users want high precision others high recall

Cleverdon’s law

Tefko Saracevic

Assumptions in Cranfield methodology IR and thus relevance is

static (traditional IR model) Relevance is:

topical binary independent stable consistent if pooling: complete

Inspired relevance experimentation on every one of these assumptions

Main finding:none of them holds

27

but simplified assumptions enabled rich IR tests and many developments

Tefko Saracevic

IR & relevance: static vs. dynamic

Q: Do relevance inferences & criteria change over time for the same user & task? A: They do For a given task, user’s inferences are dependent on

the stage of the task:Different stages = differing selections but different stages = similar criteria = different weightsIncreased focus = increased discrimination = more stringent relevance inferences

28

IR & relevance inferences are highly dynamic processes

Tefko Saracevic

Experimental results

TopicalTopicality: very important but not exclusive role.Cognitive, situational, affective variables: play a role e.g. user background (cognitive); task complexity (situational); intent, motivation (affective)

BinaryContinuum: Users judge on a continuum & comparatively, not only binary (relevant – not relevant).Bi-modality: Seems that assessments have high peaks at end points of the range (not relevant, relevant) with smaller peaks in the middle range

IndependentOrder: in which documents are presented to users seems to have an effect. Near beginning: Seems that documents presented early have a higher probability of being inferred as relevant.

29

Tefko Saracevic

Experimental results (cont.)

30

StableTime: relevance judgments = not completely stable; change over time as tasks progress & learning advancesCriteria: for judging relevance are fairly stable

ConsistentExpertise: higher = higher agreement, less differences; lower = lower agreement, more leniency. Individual differences: the most prominent feature & factor in relevance inferences. Experts agree up to 80%; others around 30%Number of judges: More judges = less agreement

If pooling:Complete

(if only a sample of collection or a pool from several searches is evaluated)Additions: with more pools or increased sampling more relevant objects are found

Tefko Saracevic

Clues: on what basis & criteria users make relevance judgments?

31

Contenttopic, quality, depth, scope, currency, treatment, clarity

Objectcharacteristics of information objects, e.g., type, organization, representation, format, availability, accessibility, costs

Validityaccuracy of information provided, authority, trustworthiness of sources, verifiability

Tefko Saracevic

Clues (cont.): Matching users

32

Use or situational match

appropriateness to situation, ortasks, usability, urgency; value in use

Cognitive match

understanding, novelty, mental effort

Affective match

emotional responses to information, fun, frustration, uncertainty

Belief match personal credence given to information,confidence

Tefko Saracevic

Summary of relevance experiments

First experiment reported in 1961 compared effects of

various representations (titles, abstracts, full text)

Over the years about 300 or so experiments

Little funding only two funded by a US

agency (1967)

33

Most important general finding:Relevance is measurable

Tefko Saracevic

In conclusion

Information technology & systems will change dramatically even in the short run and in unforeseeable directions

But relevance is here to stay!

34

and relevance has many faces – some unusual

Tefko Saracevic

Innovation ... as well ... not all are digital

35

Tefko Saracevic

and here is its use

36

Tefko Saracevic

Unusual services: Library therapy dogs

37

U Michigan, Ann Arbor, Shapiro Library

Tefko Saracevic

Presentation in Wordle

38

Tefko Saracevic 39

Thank you for inviting

me!