Query Structures

1
Query Structures
2

• Boolean Queries
• Vector Queries
• Extended Boolean Queries
• Fuzzy Queries
• Probabilistic Queries
• c.f. Natural Language Queries, DB Queries

3
Query Structures

• Query? ?? (Document?? ??)
• ??, ??(syntax) ???? ??
• ??? ??? ? ??
• Parallel Process for Matching
• Document Side
• Data? ??-gtdocument? ??(ectosystem)
• Document-gtinternal represention-gtformat for
  matching(endosystem)
• Query Side
• Information need-gtquery? ??(endosystem)
• Query-gtinternal represention-gtformat for
  matching(endosystem)

4
3.1. Matching Criteria

• Exact Match
• Numerical or business DB
• Range Match
• Exact match? ??
• Natural order(numeric or alphabetic)? ?? ??
• ??, ??? ??
• Approximate Match
• Text? image DB
• Document? query? ???? ??? ???? ??(measure)? ??
  evaluation function

5

• Exact? Approximate match? ??
• Ex) federal funding for energy development
  project, but the funding must be at least
  1,000,000

6
3.2 Boolean Queries

• Boolean Query
• Based on concepts from logic, or Boolean algebra
• (list of) Terms joined by logical
  connectives(AND, OR, NOT)
• Boolean Query? ?
• restaurants AND (Midestern OR vegetarian) AND
  inexpensive
• Expansion
• Stemming restaurant AND (Midest OR Veget) AND
  inexpens
• Thesaurus Midestern -gt list of specific countries

7
3.2 Boolean Queries

• Proximity Operator
• icing within three words of chocolate
• if icing then chocolate
• ?? ???? ??
• 2 OF (A, B, C)
• (A AND B) OR (A AND C) OR (B AND C)
• 4 OF (peony, daisy, dahlia, lily, hosta, zinnia,
  marigold)
• ?

8
3.2 Boolean Queries

• Query? ??(query? ??)
• ex) A AND B term A? ??? ?? document? ???, term
  B? ??? ?? document? ??? ???
• ????? ??
• ???? ? ??? ?? ??? ? ??
• ex) information? retrieval? ?? ???? ???
• ??? ??
• 1. information? ???? ???? D1? ???
• 2. retrieval? ???? ???? D2? ???
• 3. D1? D2? ???? ???? D3? ???

9
3.2 Boolean Queries

• ????(?? ??)
• U ?? ??
• D1, D2 ?? P1, P2? ???? ???? ??
• 1. U-D1? P1? ???? ?? ?? ??????(not)
• 2. D1nD2? P1? P2? ? ? ???? ?? ??? ????(and)
• 3. D1?D2? P1?? P2? ???? ?? ??? ????(or)
• 4. D1?D2-D1nD2? P1? P2? ????? ??? ?? ???? ?? ???
  ?? ??? ????(xor)

10
3.2 Boolean Queries

• ex)
• ??? information and retrieval or
• not retrieval and science
• ????
• (doc1,doc3 n doc1,doc2,doc4) ?
• doc1,doc2,doc3,doc4,doc5
• (doc1, doc2, doc4 n doc2, doc3,
  doc4, doc5)
• doc1 ? doc1, doc3, doc5 doc1, doc3,
  doc5

11
3.2 Boolean Queries

• ??? 1 Lack of weighting mechanism
• music by Beethoven, preferably a sonata
• Beethoven AND Sonata ???? ?? ?? ??
• Beethoven OR Sonata ?? ??? ?? ??
• (Beethoven AND Sonata) OR Beethoven
• ???? ????? Beethoven? ?? ??? ??? ??
• ??? 2 Misstated Query (and? ??)

12
3.2 Boolean Queries

• ??? 3 ?? ??
• AND, OR
• A OR A AND C
• NOTgtANDgtOR ?? strict left-to-right order
• ??? ????
• ??? ?? ??(semantic) ??? ??
• coffee AND croissant OR muffin
• raincoat AND umbrella OR sunglasses
• NOT
• ?? ?? ? ??? ????
• (NOT A) AND B AND C
• ??? B AND C? ?? ??(B AND C? ??)

13
3.2 Boolean Queries

• ??? 4 Highly Complex Query
• ??? DNF, CNF?? recast
• Disjunctive Normal Form(DNF)
• Terms ??? ??, ?? ?? ? ???
• Conjuncts AND? ?? ??? Terms
• Disjuncts OR? ?? ??? Conjuncts
• e.g. (concert AND dinner AND NOT play) OR
• (swimming AND tennis) OR
• (baseball AND NOT football)
• ?? ??? ?? query?? ?? ??? ? ??? ??? ??

14
3.2 Boolean Queries

• Full Disjunctive Normal Form
• Each conjunct contain all of the possible terms
• (A AND B) OR (A AND NOT C)
• gt (A AND B AND C) OR (A AND B AND NOT C) OR
• (A AND B AND NOT C) OR (A AND NOT B AND NOT
  C)
• Conjunctive Normal Form(CNF)
• e.g. (concert OR dinner OR NOT play) AND
• (swimming OR tennis) AND
• (baseball OR NOT football)
• Normalization
• Query? DNF? CNF? ??(transform)?? ?
• Truth table? ??
• True rows gt Full DNF

15
3.2 Boolean Queries

• Normalization? ?
• (A OR B) AND (C OR NOT D) AND (D OR B)

16
3.2 Boolean Queries

• True rows of the table
• Full DNF

17
3.2 Boolean Queries

• Minimizing to simplest possible form
• ?? ??? ?? ? ?? (A AND B AND C)? ?? ??
• ?? ? ?? ??? ???? ?? ?? ???? ??
• (A AND C AND D) OR (B AND C) OR (B AND (NOT D))
• Full CNF
• ???? false row??? full DNF? ???
• DeMorgans Law
• NOT (A AND B) (NOT A) OR (NOT B),
• NOT (A OR B) (NOT A) AND (NOT B).
• Law of Double Negation
• NOT (NOT A) A

18
3.2 Boolean Queries

• e.g. negation of query? DNF?
• (A AND B AND NOT C) OR (NOT A AND C) OR
• (B AND C) ??,
• Negation? ?? ? ???? ??? ??
• ???? (NOT A OR NOT B OR C) AND (A OR NOT C)
• AND (NOT B
  OR NOT C)

19
3.2 Boolean Queries

• ?? ??? ?? ???
• ? conjunction? ??? ??? ??? ??? ???
• A AND B? ??, ?? ??? ??? A? ???? ??? ???? ?? ???
  B? ???? ??? ???? ??
• ???, query? ? term? ???? ??? ??? ?? ? ? ???, ??
  ???? ?? ???? ? ???? ??? ???? ?? ??? ?????

20
3.2 Boolean Queries

• ??? 5 ??? ?? ??
• Query? ????? ?? document? ??? ???? ??? ??? ???
  ??? ? ??
• ???1 more restrictive query
• ???2 ?? ???? ?? (??/??? ??)
• ???? ??? sort?? ??? ??? ? ? ??
• ??? ??? ??? ?? ??
• (???? ?? sorting? ?? ??? ??? ???? ???? ???)

21
3.2 Boolean Queries

• ??
• ??? ????
• ???? ???? ???? 23 ??? ??
• ? ??? ??? ?? ?? ??? ?? ??? ???? ??? ??? ??(manual
  search??? ??)? ???? ??
• ? ??? ??? ???? ???? ??? ??
• ???? ???? ?????? ??

22
3.3 Vector Queries

• Vector Model
• Each document is represented by a vector, or
  ordered list of terms, rather than by a set of
  terms
• Boolean model?? ??
• Term representations (weights)
• Methods of determining the similarity between a
  document and the query
• Boolean model??? query? document ???? ??? ??????
  ??? ???? ???? ??

23
3.3 Vector Queries

• Similarity Evaluation 0-1 vector, weight vector
• Assigning weights to document terms in a vector
• frequency count (?? a, an, the, of, )
• user assigning
• judging dilemma freely assigned weights
• normalization

24
3.3 Vector Queries

• Retrieval Determination
• fixed number(by decreasing similarity) or
  threshold
• Impractical
• document?? components? ???? 0??. (vector?
  10000?? component(term, vocabulary)? ???? ???,
  ???? ? ? ? ??? term? ??? ? ??)
• ???
• ?? document? ??? ??? component? ??.
• ??? ???? ???? ????.
• dimensional compatibility - the comparison of
  two documents is always based on comparing the
  same terms in each document.
• Expansion of the compact representation is needed.

25
Extended Boolean Queries Boolean Query Vector
Query

• ? ??? ?? ??
• Logical connectives, weights
• Weighted Boolean query
• Boolean operation Weights(0.0 1.0)
• AW1 BW2
• Query? ?? term A? ??? ??? ?? A, term B? ??? ???
  ?? B? ???, ?? ???, ???

26
Extended Boolean Queries Boolean Query Vector
Query

• Distance
• Distance between the document sets A and B
  corresponding to the term A and B
• Minimum of the distances between a pair of
  elements
• Element a document represented by term vector
• If A contains m documents and B contains n
  documents, mn computations are needed

27
Extended Boolean Queries ??? ??

• AW1 BW2(w11, w201 )
• S weight? ?? ???? ??

28
Extended Boolean Queries ??? ?
A1,4,5,10,11,15,17,18,19,23 B1,2,4,5,7,8,10,1
3,17,23
A 0.8 OR B 0.4
w1 1
w2 1
2, 22
2, 8, 13, 22
29
Extended Boolean Queries ??? ??

• ???
• ??? ???? ??? ?? ?? 1.7, 1.4,
• ?? ??? round up or down
• ?? ??? ??? ?? ?? ??
• A? weight? 0.6?? S? ???? 4??? ?? ??? 1??? 1??
  ????? ??
• Random?? ???? ?? query? ?? ?? ??
• ????? ??? query? ?? ?? ??? ?? ? ??
• (A AND B) OR (A AND C) vs. A AND (B OR C)
• Exercise 6

30
Fuzzy Queries

• Ordinary Set vs. Fuzzy Set
• Ordinary Set sharp edge (e.g. 6feet? ???
  tall)
• Fuzzy Set membership grade
• e.g. degree of tallness? ?? membership grade
• 45 0.1, 58 0.45, 62 0,52, 610
  0.9
• Boolean operator in fuzzy set S
• query? ?? fuzzy function? ??? ?, ? document? ??
  ? ?? ??? ??.

31
Probabilistic Queries

• Fuzzy Queries
• membership grade function? 01??? ?? ???? ?? ???
  ??? ??? ? ??.
• Probabilistic Query
• the set returned from any query is supposed to
  consist of documents which satisfy that query
  with a probability higher than a specified
  threshold
• ????
• ?? ?????? ??? ??? ??? ???????

Prob(Document Satisfy)
Prob(Document not Satisfy) 1
32
Natural Language Queries

• User friendly
• Ungrammatical
• Hard to understand for computers

33
IR and DB

• Full Text retrieval System needs to combine,
• Imprecise textual element
• Precise numerical or other limit
• ?? ?? ?? ???? DB ???? ?? ??
• ? ???? ?? ??? ???? ???? ??? ??? ??? ?? ??
• One Solution OODB Model
• Object set of properties
• textual portions numeric or fixed field
  portions
• image components
• Can be Commercial???