Transcript CS411 revision class

CS 411:
Dynamic Web-Based Systems
Exam Preparation
Dr. Alexandra I. Cristea
http://www.dcs.warwick.ac.uk/~acristea/
Exam
Structure
• Time allowed: 3 hours
• This is a closed book exam. No information sources
and communication devices are allowed. Illegible
text will not be evaluated.
• Answer FOUR questions (out of SIX).
– Each 25 marks, for a total of 100 marks. This will
represent 70% of your overall mark (the rest of 30% is
coursework & presentation)
• Read carefully the instructions on the answer book
and make sure that the particulars required are
entered on each answer book.
• Day, Time, Place: 22 MAY; 09:30; Panorama Room
– Check exam time-table for changes!
2
Exam topics
1. Adaptive Hypermedia, Personalization in eCommerce
2. User Modelling
3. Authoring of Adaptive Systems, LAOS, LAG
framework, LAG language
4. Semantic Web, RDF, SPARQL, OWL
5. Social Web, Collaborative Filtering
6. Adaptive Focused Crawling, Data Mining,
Personalized Search, Privacy Enhanced Web
Personalization
3
General info
• New exam,
• But: content overlap exists with CS253
module and exam.
• Especially for topics Semantic Web,
OWL and RDF, check the old exams of
CS253.
4
1. Adaptive Hypermedia,
Personalization in e-Commerce
• Texts:
• AH: AdaptiveContentPresentation.pdf;
AdaptiveNavigationSupport.pdf;
OpenCorpusAEH.pdf; PrivacyEnhancedWebPersonalization.pdf;
UsabilityEngineeringforAdaptiveWeb.p
df
• P in eC: PersonalizationECommerce.pdf
5
1. Adaptive Hypermedia
•
•
Why, areas of application, what to adapt,
,Brusilovsky’s taxonomy, Adapt to what, (UM,
GM, DM, Envir.) how to adapt, Brusilovsky’s
loop, adaptability versus adaptivity, new
solutions.
You can be presented with a description of an
application, and asked to describe it in terms of
AH as above. E.g., what is Amazon book
recommendation adapting to? What is being
adapted? Etc.
6
1. Personalization in eCommerce
•
•
•
Benefits, perspectives, ubiquitous computing, b2b, b2c,
CRM, CDI, pull, push, generalized, personalised
recommendations, hybrid, latency (cold start), mcommerce
Again, theory and application of theory in practice; e.g., a
business personalization case is presented to you, and
you are asked to describe it in terms of the newly learned
acronyms and give the definitions. You would need to
recognize from the description which apply and which
not.
E.g., is Amazon’s book recommender technique push or
pull? Is b2b, b2c? Etc.
7
2. User Modelling
• Texts: Generic-UM.pdf; UM.pdf;
UserProfilesforPersonalizedInfoAccess.pdf
;
8
2. User Modelling
•
•
•
•
•
What, why, what for, how, early history, academic
developments, what can we adapt to (revisited, extended
– knowledge, cognitive, etc.), generic UM techniques,
new developments
Stereotypes, overlays, UM system, UM shell services +
requirements (Kobsa), semantic levels of UM, deepshallow UM, cognitive styles – Kolb, filed-dep-indep,
intended/keyhole/obstructed plan recognition, moods and
emotions, preferences
UM techniques: rule-based, frame-based, networkbased, probability, DT, sub-symbolic, example-based
Challenges for UM
UM server + requirements
9
2. User Modelling
•
Theory + application thereof either on a system you
know, or on a system with a given description; e.g., is
Amazon book recommendation based on UM shell
services, or UM server – plus justification! Or: how would
you extend the recommendation to cater for Kolb
taxonomy’s active people?
10
3. Authoring of Adaptive
Systems, LAOS, LAG
framework, LAG language
• Texts: WWWconfPaper; IFETS-journalpaper; Authoring system examples, demos
• Demos: demos (LAG, description, CAF,
AHA! demo: select anonymous session!)
11
3. Authoring of Adaptive
Systems, LAOS, LAG
framework, LAG language
•
•
•
•
What is specific to authoring of AH? Content
alternatives, UM descript, presentation,
adaptation tech., roles
LAOS components and justification,
LAG model layers and justification,
LAG language : a small program – either to read
or to write !! (based on programs you’ve been
shown, and programs you’ve been asked to
create for the coursework)
12
4. Semantic Web, RDF, SPARQL,
OWL
• Texts: READING GUIDE; SW: SPARQL (to
be read online);
online testing
• Some extra courses to visit:
– RDF course ; video;
– OWL course ; video;
– SPARQL course ; video;
13
4. Semantic Web, RDF, SPARQL,
OWL
• SW: inventor, sytactic vs SW, ontology
def., SW ontology languages, ‘Layer Cake’
14
4. Semantic Web, RDF, SPARQL,
OWL
• RDF: def, purpose, syntax, graphical and
RDF/XML representations – you should be
able to represent your data in RDF;
namespaces – why and how in RDF/XML,
resource, description, properties as
attributes, resources, elements, containers
– bag, seq, alt -, collections, reification,
RDF Schema – classes, subclasses (long,
short-hand notation -), range, domain, type
15
4. Semantic Web, RDF, SPARQL,
OWL
•
OWL: def, purpose, sublanguages, individuals,
object properties (domain, range from RDF),
restrictions on prop. (allValuesFrom,
someValuesFrom, hasValue, minCardinality,
maxCardinality, cardinality), inverse prop., trans.
Prop., sub-prop., datatype prop., owl classes –
disjoint, enumerated classes - oneOf,
intersectionOf, complementOf, unionOf, class
Conditions – necessary, nec+suff., reasoning,
ontology extension,
16
4. Semantic Web, RDF, SPARQL,
OWL
• SPARQL: what for?; SELECT,
CONSTRUCT, ASK, DESCRIBE (you
should be able to know the difference
between them, and to read some simple
queries, mainly based on SELECT)
17
5. Social Web, Collaborative
Filtering
• Texts: RecommendationGroups.pdf;
AdaptiveSupportDistributedCollaboration.p
df; HybridWebRecommenderSystems.pdf ;
CollaborativeFiltering.pdf
18
5. Social Web, Collaborative
Filtering
•
•
Web 2.0, user profiling (explicit-implicit data
collection), content-based filtering (items,
grouping, rating, accuracy), collaborative filtering
(automatic; rating patterns; sharing; advantages
– disadvantages; passive-active; explicit-implicit;
first-rater; cold-start), hybrid filtering, group
recommendations, social filtering (similarity
computations)
You can be asked theory questions, you can be
asked to discuss the topics, you can be asked
how a given system fairs in term of the theory
you’ve learned
19
6. Adaptive Focused Crawling,
Data Mining, Personalized
Search, Privacy Enhanced Web
Personalization
•
•
These are topics based on the last topic,
crawling, and your presentations. grouped
together. Your main source for the group
presentations should be the text (literature).
Texts: AdaptiveFocusedCrawling.pdf ;
DataMining.pdf ; PersonalizedSearch.pdf;
Privacy-EnhancedWebPersonalization.pdf
20
6. Adaptive Focused Crawling,
Data Mining, Personalized
Search, Privacy Enhanced Web
Personalization
•
•
•
Crawling: on the WWW, focused c. (adaptive or
not; dark matter, page sets: In, Out, SCC, deep
web; strategies – BF, Backlink, PageRank, HITS,
fish, tunneling, etc.), agent-based (genetic, ants),
ML (statistical model), eval. Methods (time,
precision, recall)
Theory + discussion & interpretation
Small problems/ numerical computations based
on theory
21
6. Adaptive Focused Crawling,
Data Mining, Personalized
Search, Privacy Enhanced Web
Personalization
•
•
Data mining: def, cycle, collection,
preprocessing (+ tasks, web-usage, fusion,
cleaning, pageview identification, sessionization,
episode id, ), modelling (offline, clustering, rule
discovery, sequential models, LVM; hybrids),
representation, data sources, recommendations,
evaluations
Theory + discussion & interpretation
22
6. Adaptive Focused Crawling,
Data Mining, Personalized
Search, Privacy Enhanced Web
Personalization
•
•
Personalised Search: def, surf, query,
content/collaborative-based (polysemy, synonymy), user
modeling, profiling, re-ranking, query modification,
relevance feedback, query expansion, contextualised,
search histories, agents, offline-online, rich
representations (frames, AI, UM, stereotypes, feedback),
collaborative search (similarity, statistics, communities),
adaptive result clustering, hyperlink-based
personalisation, combined approaches
Theory + discussion & interpretation
23
6. Adaptive Focused Crawling,
Data Mining, Personalized
Search, Privacy Enhanced Web
Personalization
•
•
Privacy-enhanced Web personalisation: concerns
(personalisation vs. privacy; methods, effects,
differences), factors (knowledge, trust, benefits, costs,
hyperbolic temporal discounting, ), laws (on what?; EU?;
ACM list of recommendations), technology
(pseudonymous, anonymous, client-side, centralised,
issues, perturbation/ obfuscation, personalising privacy)
Theory + discussion & interpretation
24
• Questions?
25