Transcript Modeling and Understanding Human Behavior on the Web

Modeling the Internet and the
Web:
Modeling and Understanding
Human Behavior on the Web
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
Outline
• Introduction
• Web Data and Measurement Issues
• Empirical Client-Side Studies of Browsing
Behavior
• Probabilistic Models of Browsing Behavior
• Modeling and Understanding Search
Engine Querying
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
2
Introduction
• Useful to study human digital behavior,
e.g. search engine data can be used for
– Exploration e.g. # of queries per session?
– Modeling e.g. any time of day dependence?
– Prediction e.g. which pages are relevant?
• Helps
– Understand social implications of Web usage
– Better design tools for information access
– In networking, e-commerce etc
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
3
Web data and measurement issues
Background:
• Important to understand how data is collected
• Web data is collected automatically via software
logging tools
– Advantage:
• No manual supervision required
– Disadvantage:
• Data can be skewed (e.g. due to the presence of robot traffic)
• Important to identify robots (also known as
crawlers, spiders)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
4
A time-series plot of Web requests
Number of page requests per hour as a function of time from page
requests in the www.ics.uci.edu Web server logs during the first week of
April 2002.
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
5
Robot / human identification
• Robot requests identified by classifying
page requests using a variety of heuristics
– e.g. some robots self-identify themselves in
the server logs (robots.txt)
– Robots explore the entire website in breadth
first fashion
– Humans access web-pages in depth first
fashion
• Tan and Kumar (2002) discuss more
techniques
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
6
Robot / human identification
• Robot traffic consists of two components
– Periodic Spikes (can overload a server)
• Requests by “bad” robots
– Lower-level constant stream of requests
• Requests by “good” robots
• Human traffic has
– Daily pattern: Monday to Friday
– Hourly pattern: peak around midday & low
traffic from midnight to early morning
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
7
Server-side data
Data logging at Web servers
• Web server sends requested pages to the
requester browser
• It can be configured to archive these
requests in a log file recording
– URL of the page requested
– Time and date of the request
– IP address of the requester
– Requester browser information (agent)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
8
Data logging at Web servers
– Status of the request
– Referrer page URL if applicable
• Server-side log files
– provide a wealth of information
– require considerable care in interpretation
• More information in Cooley et al. (1999),
Mena (1999) and Shahabi et al. (2001)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
9
Page requests, caching, and proxy
servers
• In theory, requester browser requests a
page from a Web server and the request is
processed
• In practice, there are
– Other users
– Browser caching
– Dynamic addressing in local network
– Proxy Server caching
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
10
Page requests, caching, and proxy
servers
A graphical summary of how page requests from an individual user can be
masked at various stages between the user’s local computer and the Web
server.
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
11
Page requests, caching, and proxy
servers
• Web server logs are therefore not so ideal in
terms of a complete and faithful representation
of individual page views
• There are heuristics to try to infer the true
actions of the user: – Path completion (Cooley et al. 1999)
• e.g. If known B -> F and not C -> F, then session ABCF can
be interpreted as ABCBF
• Anderson et al. 2001 for more heuristics
• In general case, hard to know what user viewed
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
12
Identifying individual users from
Web server logs
• Useful to associate specific page requests to
specific individual users
• IP address most frequently used
• Disadvantages
– One IP address can belong to several users
– Dynamic allocation of IP address
• Better to use cookies
– Information in the cookie can be accessed by the
Web server to identify an individual user over time
– Actions by the same user during different sessions
can be linked together
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
13
Identifying individual users from
Web server logs
• Commercial websites use cookies extensively
• 90% of users have cookies enabled permanently
on their browsers
• However …
– There are privacy issues – need implicit user
cooperation
– Cookies can be deleted / disabled
• Another option is to enforce user registration
– High reliability
– Can discourage potential visitors
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
14
Client-side data
• Advantages of collecting data at the client side:
– Direct recording of page requests (eliminates ‘masking’ due to
caching)
– Recording of all browser-related actions by a user (including
visits to multiple websites)
– More-reliable identification of individual users (e.g. by login ID for
multiple users on a single computer)
• Preferred mode of data collection for studies of
navigation behavior on the Web
• Companies like comScore and Nielsen use client-side
software to track home computer users
• Zhu, Greiner and Häubl (2003) used client-side data
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
15
Client-side data
• Statistics like ‘Time per session’ and ‘Page-view
duration’ are more reliable in client-side data
• Some limitations
– Still some statistics like ‘Page-view duration’ cannot
be totally reliable e.g. user might go to fetch coffee
– Need explicit user cooperation
– Typically recorded on home computers – may not
reflect a complete picture of Web browsing behavior
• Web surfing data can be collected at
intermediate points like ISPs, proxy servers
– Can be used to create user profile and target
advertise
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
16
Handling massive Web server logs
• Web server logs can be very large
– Small university department website gets a million
requests per month
– Amazon, Google can get tens of millions of requests
each day
• Exceed main memory capacities, stored on
disks
• Time-costs to data access place significant
constraints on types of analysis
• In practice
– Analysis of subset of data
– Filtering out events and fields of no direct interest
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
17
Empirical client-side studies of
browsing behavior
• Data for client-side studies are collected at the
client-side over a period of time
–
–
–
–
–
Reliable page revisitation patterns can be gathered
Explicit user permission is required
Typically conducted at universities
Number of individuals is small
Can introduce bias because of the nature of the
population being studied
– Caution must be exercised when generalizing
observations
• Nevertheless, provide good data for studying
human behavior
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
18
Early studies from 1995 to 1997
• Earliest studies on client-side data are Catledge
and Pitkow (1995) and Tauscher and Greenberg
(1997)
• In both studies, data was collected by logging
Web browser commands
• Population consisted of faculty, staff and
students
• Both studies found
– clicking on the hypertext anchors as the most
common action
– using ‘back button’ was the second common action
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
19
Early studies from 1995 to 1997
– high probability of page revisitation (~0.58-0.61)
• Lower bound because the page requests prior to the start of
the studies are not accounted for
• Humans are creatures of habit?
• Content of the pages changed over time?
– strong recency (page that is revisited is usually the
page that was visited in the recent past) effect
• Correlates with the ‘back button’ usage
• Similar repetitive actions are found in telephone
number dialing etc
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
20
The Cockburn and McKenzie study
from 2002
• Previous studies are relatively old
• Web has changed dramatically in the past few
years
• Cockburn and McKenzie (2002) provides a more
up-to-date analysis
– Analyzed the daily history.dat files produced by the
Netscape browser for 17 users for about 4 months
– Population studied consisted of faculty, staff and
graduate students
• Study found revisitation rates higher than past
94 and 95 studies (~0.81)
– Time-window is three times that of past studies
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
21
The Cockburn and McKenzie study
from 2002
• Revisitation rate less biased than the previous studies?
• Human behavior changed from an exploratory mode to a
utilitarian mode?
– The more pages user visits, the more are the
requests for new pages
– The most frequently requested page for each user
can account for a relatively large fraction of his/her
page requests
• Useful to see the scatter plot of the distinct
number of pages requested per user versus the
total pages requested
• Log-log plot also informative
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
22
The Cockburn and McKenzie study
from 2002
The number of distinct pages visited versus page vocabulary size of each
of the 17 users in the Cockburn and McKenzie (2002) study
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
23
The Cockburn and McKenzie study
from 2002
The number of distinct pages visited versus page vocabulary size of each
of the 17 users in the Cockburn and McKenzie (2002) study (log-log plot)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
24
The Cockburn and McKenzie study
from 2002
Bar chart of the ratio of the number of page requests for the most frequent
page divided by the total number of page requests, for 17 users in the
Cockburn McKenzie (2002) study
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
25
Video-based analysis of Web
usage
• Byrne et al. (1999) analyzed video-taped
recordings of eight different users over a period
of 15 min to 1 hour
• Audio descriptions of the users was combined
with the video recordings of their screen for
analysis
• Study found
– users spent a considerable amount of time scrolling
Web pages
– users spent a considerable amount of time waiting for
pages to load (~15% of time)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
26
Probabilistic models of browsing
behavior
• Useful to build models that describe the
browsing behavior of users
• Can generate insight into how we use
Web
• Provide mechanism for making predictions
• Can help in pre-fetching and
personalization
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
27
Markov models for page prediction
• General approach is to use a finite-state Markov
chain
– Each state can be a specific Web page or a category
of Web pages
– If only interested in the order of visits (and not in
time), each new request can be modeled as a
transition of states
• Issues
– Self-transition
– Time-independence
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
28
Markov models for page prediction
• For simplicity, consider order-dependent, timeindependent finite-state Markov chain with M states
• Let s be a sequence of observed states of length L. e.g.
s = ABBCAABBCCBBAA with three states A, B and C. st
is state at position t (1<=t<=L). In general,
L
P( s)  P( s1 ) P( st | st 1 ,...,s1 )
t 2
• Under a first-order Markov assumption, we have
L
P( s)  P( s1 ) P( st | st 1 )
t 2
• This provides a simple generative model to produce
sequential data
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
29
Markov models for page prediction
• If we denote Tij = P(st = j|st-1 = i), we can define a M x M
transition matrix
• Properties
– Strong first-order assumption
– Simple way to capture sequential dependence
• If each page is a state and if W pages, O(W2), W can be
of the order 105 to 106 for a CS dept. of a university
• To alleviate, we can cluster W pages into M clusters,
each assigned a state in the Markov model
• Clustering can be done manually, based on directory
structure on the Web server, or automatic clustering
using clustering techniques
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
30
Markov models for page prediction
• Tij = P(st = j|st-1 = i) now represent the probability
that an individual user’s next request will be from
category j, given they were in category i
• We can add E, an end-state to the model
• E.g. for three categories with end state:  P(1 | 1) P(2 | 1) P(3 | 1 _) P( E | 1) 


 P(1 | 2) P(2 | 2) P(3 | 2) P( E | 2) 
T 
P(1 | 3) P(2 | 3) P(3 | 3) P( E | 3) 


 P(1 | E ) P(2 | E ) P(3 | E )
0 

• E denotes the end of a sequence, and start of a
new sequence
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
31
Markov models for page prediction
• First-order Markov model assumes that the next
state is based only on the current state
• Limitations
– Doesn’t consider ‘long-term memory’
• We can try to capture more memory with kthorder Markov chain
P(st | st 1 ,.., s1 )  P(st | st 1 ,.., st k )
• Limitations
– Inordinate amount of training data O(Mk+1)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
32
Fitting Markov models to observed
page-request data
• Assume that we collected data in the form of N
sessions from server-side logs, where ith session
si, 1<= i <= N, consists of a sequence of Li page
requests, categorized into M – 1 states and
terminating in E. Therefore, data D = {s1, …, sN}
• Let  denote the set of parameters of the Markov
model,  consists of M2 -1 entries in T
• Let ij denote the estimated probability of
transitioning from state i to j.
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
33
Fitting Markov models to observed
page-request data
• The likelihood function would be
N
L( )  P( D |  )   P( si |  )
i 1
• This assumes conditional independence of sessions.
• Under Markov assumptions, likelihood is
L( )   ijij ,1  i, j  M
n
• where nij is the number of times we see a transition from
state i to state j in the observed data D.
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
34
Fitting Markov models to observed
page-request data
• For convenience, we use log-likelihood
l ( )  log L( )   nij log ij
ij
• We can maximize the expression by taking
partial derivatives wrt each parameter and
incorporating the constraint (via Lagrange
multipliers) that the sum of transition
probabilities out of any state must sum to one

ij
1
j
• The maximum likelihood (ML) solution is 
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
ML
ij

nij
ni
35
Bayesian parameter estimation for
Markov models
• In practice, M is large (~102-103), we end up estimating
M2 probabilities
• D may contain potentially millions of sequences, so
some nij = 0
• Better way would be to incorporate prior knowledge –
prior probability distribution P( ) and then maximize
, the posterior distribution on P( | D) given the data (rather
than P( D |  ) )
• Prior distribution reflects our prior belief about the
parameter set 
• The posterior reflects our posterior belief in the
parameter set now informed by the data D
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
36
Bayesian parameter estimation for
Markov models
• For Markov transition matrices, it is common to put a
distribution on each row of T and assume that each of
these priors are independent
P( )   P({ ,..., })
where   1
• Consider the set of parameters for the ith row in T, a
useful prior distribution on these parameters is the
Dirichlet distribution defined as
i1
iM
i
ij
j
M
P({ i1 ,..., iM })  Dqi  C ij
 ( qij 1)
j 1
• where
 , qij  0,  qij  1
, and C is a normalizing constant
j
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
37
Bayesian parameter estimation for
Markov models
• The MP posterior parameter estimates are

MP
ij

nij  qij
ni  
• If nij = 0 for some transition (i, j) then instead of
having a parameter estimate of 0 (ML), we will
have q /(n   ) allowing prior knowledge to be
incorporated
• If nij > 0, we get a smooth combination of the
data-driven information (nij) and the prior
ij
i
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
38
Bayesian parameter estimation for
Markov models
• One simple way to set prior parameter is
– Consider alpha as the effective sample size
– Partition the states into two sets, set 1 containing all
states directly linked to state i and the remaining in
set 2
– Assign uniform probability e/K to all states in set 2 (all
set 2 states are equally likely)
– The remaining (1-e) can be either uniformly assigned
among set 1 elements or weighted by some measure
– Prior probabilities in and out of E can be set based on
our prior knowledge of how likely we think a user is to
exit the site from a particular state
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
39
Predicting page requests with
Markov models
• Many flavors of Markov models proposed for
next page and future page prediction
• Useful in pre-fetching, caching and
personalization of Web page
• For a typical website, the number of pages is
large – Clustering is useful in this case
• First-order Markov models are found to be
inferior to other types of Markov models
• kth-order is an obvious extension
– Limitation: O(Mk+1) parameters (combinatorial
explosion)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
40
Predicting page requests with
Markov models
• Deshpande and Karypis (2001) propose
schemes to prune kth-order Markov state space
– Provide systematic but modest improvements
• Another way is to use empirical smoothing
techniques that combine different models from
order 1 to order k (Chen and Goodman 1996)
• Cadez et al. (2003) and Sen and Hansen (2003)
propose mixtures of Markov chains, where we
replace the first-order Markov chain:
P(st | st 1 ,...,s1 )  P(st | st 1 )
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
41
Predicting page requests with
Markov models
with a mixture of first-order Markov chains
K
P( st | st 1 ,...,s1 )   P( st | st 1 , c  k )P(c  k )
k 1
• where c is a discrete-value hidden variable taking K
values Sumk P(c = k) = 1and
P(st | st-1, c = k) is the transition matrix for the kth mixture
component
• One interpretation of this is user behavior consists of K
different navigation behaviors described by the K Markov
chains
• Cadez et al. use this model to cluster sequences of page
requests into K groups, parameters are learned using
the EM algorithm
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
42
Predicting page requests with
Markov models
• Consider the problem of predicting the next state, given
some number of states t
• Let s[1,t] = {s1,…, st} denote the sequence of t states
• The predictive distribution for a mixture of K Markov
models is
K
P( st 1 | s[1,t ] )   P( st 1 , c  k | s[1,t ] )
k 1
K
  P( st 1 | s[1,t ] , c  k ) P(c  k | s[1,t ] )
k 1
K
  P( st 1 | st , c  k ) P(c  k | s[1,t ] )
k 1
• The last line is obtained if we assume conditioned on component c =
k, the next state st+1 depends only on st
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
43
Predicting page requests with
Markov models
• Weight based on observed history is
P(c  k | s[1,t ] ) 
P( s[1,t ] | c  k ) P(c  k )
 P( s
[1,t ]
| c  j ) P(c  j )
,1  k  K
j
L
where P(s | c  k )  P(s1 | c  k ) P(s | s , c  k )
• Intuitively, these membership weights ‘evolve’ as
we see more data from the user
• In practice,
[1,t ]
t
t 1
t 2
– Sequences are short
– Not realistic to assume that observed data is
generated by a mixture of K first-order Markov chains
• Still, mixture model is a useful approximation
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
44
Predicting page requests with
Markov models
• K can be chosen by evaluating the out-ofsample predictive performance based on
– Accuracy of prediction
– Log probability score
– Entropy
• Other variations of Markov models
– Sen and Hansen 2003
– Position-dependent Markov models (Anderson et al.
2001, 2002)
– Zukerman et al. 1999
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
45
Search Engine Querying
• How users issue queries to search engines
– Tracking search query logs
timestamp, text string, user ID etc.
– Collecting query datasets from different distribution
Jansen et al (1998), Silverstein et al (1998)
Lau and Horvitz (1999), Spink et al (2002)
Xie and O’Hallaron (2002)
e.g.
Xie and O’Hallaron (2002)
• Checked how many queries were coming
• Checked “user’s” IP address
• Reported 111,000 queries (2.7%) originating from AOL
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
46
Analysis of Search Engine
Query Logs
# of Sample Query Source SE
Time Period
Lau & Horvitz
4690 of 1 Million
Excite
Sep 1997
Silverstein et al
1 Billion
AltaVista
6 weeks in Aug &
Sep 1998
Spink et al
(series of studies)
1Million for each
time period
Excite
Sep 1997
Dec 1999
May 2001
Xie &
O’Hallaron
110,000
Vivisimo
35 days Jan & Feb
2001
1.9 Million
Excite
8 hrs in a day, Dec
1999
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
47
Main Results
• Average number of terms in a query is ranging
from a low of 2.2 to a high of 2.6
• The most common number of terms in a query is
2
• The majority of users don’t refine their query
– The number of users who viewed only a single page
increase 29% (1997) to 51% (2001) (Excite)
– 85% of users viewed only first page of search results
(AltaVista)
• 45% (2001) of queries is about Commerce,
Travel, Economy, People (was 20%1997)
– The queries about adult or entertainment decreased
from 20% (1997) to around 7% (2001)
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
48
Main Results
- Query Length
Distributions (bar)
- Poisson Model
(dots & lines)
• All four studies produced a generally consistent
set of findings about user behavior in a search
engine context
– most users view relatively few pages per query
– most users don’t use advanced search features
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
49
Advanced Search Tips
• Useful operators for searching (Google)
+
Include stop word (common words)
+where +is Irvine
Exclude
operating system -Microsoft
~
Synonyms
~computer
“…“ Phrase search
“modeling the internet”
or
Either A Or B
vacation London or Paris
site: Domain search
admission site:www.uci.edu
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
50
Power-law Characteristics
Power-Law
in log-log space
• Frequency f(r) of Queries with Rank r
– 110000 queries from Vivisimo
– 1.9 Million queries from Excite
• There are strong regularities in terms of patterns of
behavior in how we search the Web
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
51
Models for Search Strategies
• It is significant to know the process by which a
typical user navigates through search space
when looking for information using a search
engine
• The inference of user’s search actions could be
used for marketing purposes such as real-time
targeted advertising
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
52
Graphical Representation
•
Lar & Horvitz(1999)
–
–
Model of user’s search query actions over time
Simple Bayesian network
1)
2)
3)
4)
Current search action
Time interval
Next search action
Informational goals
–
Track ‘Search Trajectory’ of individual users
–
Provide more relevant feedback to users
Modeling the Internet and the Web
School of Information and Computer Science
University of California, Irvine
53