Download F. Fakour

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Beyond Basic Faceted Search
Ben-Yitzhak, et al.
Fahimeh Fakour
CS 572
Summer 2010
Introduction
1.
2.
3.
4.
5.
6.
7.
8.
7/7/2010
Importance and significance
Background Information
Objective
Related work
Approach and Solutions
Enhancements
Contributions
Pros & Cons
Beyond Basic Faceted Search
2
1. Importance and
Significance
• Too much info
• Transactions
7/7/2010
Beyond Basic Faceted Search
3
1. Importance and
Significance (cont)
• Categories, lists, and the human
mind
7/7/2010
Beyond Basic Faceted Search
4
2. Background Information
• Research done in IBM & Yahoo
Research labs
• Facets, buckets, and categories
– Navigate multiple paths for different
ordering
• Free text queries
• List of matching
documents with
count
7/7/2010
Beyond Basic Faceted Search
5
3. Objective
• Extend traditional facet
– Beyond numbers
Numbers
Words
• Search & Index correlated
documents
• Similarity to OLAP: multidimensional data
7/7/2010
Beyond Basic Faceted Search
6
4. Related Work
• Multifaceted search
– Lexical subsumption
– Synsets and hypernym
– RawSugar social tagging
• Online Analytical Processing (OLAP)
– Multi-dimensional data
– Aggregation of data: Cube
• N-dimensional “group by”
Exciting new technique
7/7/2010
Beyond Basic Faceted Search
7
5. Approach & Solutions
5.1
5.2
5.3
5.4
5.5
5.6
7/7/2010
Technologies: Lucene & Solr
Data Model
Facet hierarchy: Forest
Creating the facet paths
Running the facet query
Example
Beyond Basic Faceted Search
8
5.1. Technologies:
Lucene & Solr
• Posting element:
byte array of additional
info (runtime accessible)
docID, offset, payload
• Matching document processing
7/7/2010
Beyond Basic Faceted Search
9
5.2. Data Model
• Taxonomy: hierarchical
relationships among facets
– Predefined taxonomy
– Acquired/Learned through
documents

• Facet-path forest
– Tree: top-level facet
7/7/2010
Beyond Basic Faceted Search
10
5.3. Facet hierarchy: Forest
Find facet
hierarchies
Map documents
to that
hierarchy
7/7/2010
Beyond Basic Faceted Search
11
5.4. Creating the facet
paths
• Posting element for document for
each prefix of Pi
• Add path to taxonomy index
• Encode all k paths related to this
document
7/7/2010
Beyond Basic Faceted Search
12
5.5. Running the facet
query
• Terms:
– Faceted query string + taxonomy
subtrees
– Faceted result set  ranked list of
documents matching query + counters
• Lucene: use the Taxonomy Index
function to determine ordinal
number of paths
7/7/2010
Beyond Basic Faceted Search
13
5.6. Example
Clothing
Winter
Coats
Children’s
Coats
Color
Red
$30-$40
7/7/2010
$36-$40
All seasons
Accessories
Price
$30-$35
Women’s
Facet$clothing:
Facet$clothing$children’s:
Beyond Basic Faceted Search
Blue
doc1,doc2
doc1
14
6. Enhancements
7/7/2010
Beyond Basic Faceted Search
15
6.1. Business Intelligence
• Qualitative rather than quantitative
– Best sellers rather than number of
books published by author
7/7/2010
Beyond Basic Faceted Search
16
6.2. Dynamic Facets:
Welcome to the real world
• Not always independent data
• Example:
– Running shorts
• Different sizes per color
• Location & price
7/7/2010
Beyond Basic Faceted Search
17
6.2. Dynamic Facets:
Solution
• Use tree over the data
Manufacturer: Arthur’s Sports
Model: Excalibur
Type: Running Shorts
Color: red
7/7/2010
Color: blue
Color: black
Size: small
Size: medium
Store: NY
Store: SJ
Price: $20
Price: $15
Beyond Basic Faceted Search
Store: NY
Price: $20
Store: SJ
Price: $15
18
6.2. Dynamic Facets:
Solution (cont)
Manufacturer: Arthur’s Sports
Model: Galahad
Type: Running Shorts
Store: SJ
7/7/2010
Color: blue
Color: black, white
Size: small
Size: medium, large
Price: $20
Price: $12
Beyond Basic Faceted Search
19
7. Contributions
• “rich” aggregation : qualitative
• Engineering details
• Correlation in facet values
7/7/2010
Beyond Basic Faceted Search
20
8.1.  Pros 
• Detailed description of engineering
aspects & design decisions
• Use of implemented technologies
• Clearly defines the scope of the
paper
• Give foundation/background
information
• Compatible with real life data
7/7/2010
Beyond Basic Faceted Search
21
8.2.  Cons 
• Experiments and testing: No
qualitative measurement
– effectiveness of “qualitative” facets
• Not explain relevance of some of
the previous work
• Criteria for display/grouping?
– Key use cases & known user access
patterns not explained
• Build taxonomy: depth/breadth?
7/7/2010
Beyond Basic Faceted Search
22
Thank You
7/7/2010
Beyond Basic Faceted Search
23
References
Ben-Yitzhak, et al. “Beyond Basic Faceted Search”.
Proceedings of the international conference on Web
search and web data mining. Pp.33-44, 2008.
<http://nadav.harel.org.il/papers/p33-ben-yitzhak.pdf>
“Faceted Search with Solr” Lucid Imagination. July 1, 2010.
<http://www.lucidimagination.com/Community/Hearfrom-the-Experts/Articles/Faceted-Search-Solr >
“Faceted classification” Wikipedia. July 7, 2010
<http://en.wikipedia.org/wiki/Faceted_classification >
Lemieux, Earley, and Associates. “Designing for Faceted
Search” User Interface Engineering. July 6, 2010
<http://www.uie.com/articles/faceted_search/>
(Originally in KM World, March 2009)
Mattman, Chris. “Query Models” (presentation slides for
class)
7/7/2010
Beyond Basic Faceted Search
24