Skip to main content

Mapping Knowledge Domains (2004) / Chapter Skim
Currently Skimming:

An unsupervised method for the extraction of propositional information from text
Pages 24-31

The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.


From page 24...
... The majority of current information extraction systems are based on surface analysis of text applied to very large textbases. Whereas the dominant approaches in the late 1980s and early 1990s would attempt deep linguistic analysis, proposition extraction, and reasoning, most current systems look for answer patterns within the raw text and apply simple heuristics to extract relevant information (3~.
Read the entire page →
From page 25...
... Sampras defeated Agassi ~ ~ ~ [A1] Kuerten defeated Roddick In this alignment, we have three edit operations.
Read the entire page →
From page 26...
... 5208 1 www.pnas.org/cgi/doi/10.1073/pnas.0307758101 ? # Working Memory Who IWho 1 ~ did | did Sampras Sampras beat | beat l L ~ 1# 1 Sequential LTM Sampras defeated Agassi Kuerten defeated Roddick Hewitt defeated Costa Who did Kuerten beat?
Read the entire page →
From page 27...
... Roddick tale rtird F4P`A/~1t beats Cents Relational Retrieval. The bindings of input words to their corresponding role vectors (the relational representation of the target sentence)
Read the entire page →
From page 28...
... Roddick Who did Hewitt beat? Costa Sequential Relational LTM Samp~ras: Kuerten Hewitt; Agassi: Roddick, Costa ts~?
Read the entire page →
From page 29...
... First, the series of words in the target sentence is used to retrieve traces that are similar from sequential LTM. Then, the retrieved sequential traces are aligned with the input sentence to create a relational interpretation of the sentence based on word order.
Read the entire page →
From page 30...
... 7. Breakdown of responses based on the accuracy of the response and the type of the most probable relational trace according tothe model.
Read the entire page →
From page 31...
... Etlis," it returned "Kiefer quickly overcame Gaston Etlis of Argentina 6-2, 6-4 on Monday to qualify for the main draw of the Kremlin cup." Fi~ 7 shows the numher of most prohahle relational traces in For an indication of the contnbution that inference by coincidence is making to correct responding, consider those correct responses that can be attnbuted to either literal or inference traces. On 59% of occasions, the model was infernng the answer rather than rely~ng on literal retrieval.
Read the entire page →

Key Terms

  • target sentence
  • defeated agassi
  • relational trace
  • defeated roddick
  • kuerten defeated

  • edit operations
  • relational ltm
  • information extraction
  • role vector
  • relevant information

  • retrieval probability
  • propositional information
  • tennis news
  • sequential ltm
  • edit operation

  • retrieval conference
  • probable relational
  • text retrieval
  • match probabilities
  • edit theory

  • roddick defeated
  • relational retrieval
  • hewitt defeated
  • difficult task
  • edit probabilities

  • sampras beat
  • sampras outguns
  • string edit
  • log likelihood
  • kuerten beat


    • This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
    More information on Chapter Skim is available.