Distributed Information Retrieval

1
Distributed Information Retrieval

• Server Ranking for Distributed Text Retrieval
  Systems on the Internet
• B. Yuwono and D. Lee
• Siemens TREC-4 Report Further Experiments with
  Database Merging E. Vorhees

Brian Shaw CS 5604
2
Issue Merging for Effective Results

• multiple brokers (take search queries), multiple
  collection servers
• broker must select appropriate collection servers
  and merge results

3
Server Ranking overview

• Problem cost (including users time) of
  broadcasting to all servers and processing power
• Solution broker ranks collection servers
  (goodness score) broadcasts query to at most s
  (sigma) collection servers (preset number or
  scoring threshold) merges results

1- Server Ranking for Distributed Text Retrieval
on the Internet
4
Server Ranking Server Selection

• Relies solely on Document Frequency data (DF)
  all collection servers must report changes to
  broker
• Cue Validity Variance (CVV) goodness score is
  based on estimate that term j distinguishes one
  collection server from another not an indication
  of quantity or quality of relevance

1- Server Ranking for Distributed Text Retrieval
on the Internet
5
Server Ranking Merging

• Assumption 1 the best document in collection i
  is equally relevant to the best document in
  collection k
• A collection server containing a few but highly
  relevant documents will contribute to the final
  list.
• Assumption 2 the distance between two
  consecutive document ranks is inversely
  proportional to the goodness score
• Relative goodness scores are roughly proportional
  to the number of documents contributed to the
  final list.
• Final ranking is a combination of goodness score
  and local rankings.

1- Server Ranking for Distributed Text Retrieval
on the Internet
6
Experiments (overview)

• Problem broker has no access to meta-data from
  isolated collection servers
• Solution choose collection server(s) based on
  results from previous training queries

2- Further Experiments with Database Merging
7
Experiments Server Selection, two approaches

• Query Clustering (QC) cluster training queries
  (based on of same documents retrieved) and
  calculate cluster centroid vector compare
  query vector to centroid vector and assign weight
  to collection
• Modeling Relevant Document Distributions (MRDD)
  find M most similar training queries and assign
  weights to collections based on the training
  runs relevant document distribution

2- Further Experiments with Database Merging
8
Experiments Merging

• N documents retrieved from each server as
  determined by weights
• Final ranking is a random process roll a C-faced
  die that is biased by the number of documents
  still to be picked from each of the C collections

2- Further Experiments with Database Merging
9
Comparison
1-Server Ranking 2-Experiments
Brokers Knowledge Shared Document Frequency Data Training Query Results
Collection Server Selection CVV Goodness Scoring Comparison to Training Queries
Merging Goodness Score Local Rank Random
10
Conclusions

• The server ranking method proposed by Yuwono and
  Lee is an effective way to minimize operating
  costs (such as time) in an environment where
  brokers and collection servers can share document
  frequency data.
• The isolated merging strategies proposed by
  Vorhees is an effective way to choose a
  collection server where no meta-information is
  shared between the broker and collection server.