ANNIS: Search and Visualization in Multilayer Linguistic Corpora

New: version 3.1.6 has been released (github log) - please install the new version

New features in version 3.1.6

• Note: this is a bug fix version addressing minor issues in 3.1.4. The features below are available from 3.1.4.
• Backend (database):
• Support for aggregate queries
• Fetching of full text for documents
• Storage for individual corpus preferences
• Minor performance improvements


• Frontend (user interface):
• New frequency analysis interface (histograms and type counts based on combinations of query elements and their annotations)
• Document browser mode for close reading
• Get more/less context for individual search results
• Re-implemented key-word-in-context (KWIC) for better support of dialogue corpora and gaps
• Better handling of ‘islands’: search results containing very distant areas of a document hide intervening text more intuitively for the grid and KWIC visualizers


• Query language:
• Shortened query form with operators between elements:
  (e.g. cat="NP" > cat="PP" or "hello" . "world")
• New non-binding value comparison operators:
  tok . tok & #1 == #2 (finds a sequence of two identical tokens)
  tok _=_ lemma & #1 != #2 (finds a token that is not identical to its lemma)
  
• Free naming of query nodes, e.g.
  NP#cat="NP" & PP1#cat="PP" . PP2#cat="PP" & #NP > #PP1 & #NP > #PP2
• Extended support for brackets in disjunctions, e.g.:
  cat="NP" & (head#cat="NP" | head#pos="NN") & #1 >[func="HD"] #head

(For change logs of previous versions see their respective distributions or user guides)

What's new in ANNIS since ANNIS2?

Some of the new features in ANNIS3 include:

• Backend (database):
  • Support for multiple/overlapping tokenizations (multiple simultaneous speakers, conflicting tokenizations produced by different tools)
  • Support for subtokenization (annotations smaller than the reference word form unit)
  • Incremental search result fetching


• Frontend (user interface):
  • Completely new front-end architecture (using VAADIN)
  • Build your own HTML visualization with CSS3
  • Support for time-aligned streaming A/V data
  • On click navigation from annotations to time-aligned player output
  • Page-aligned PDF-viewer
  • View KWIC and context using multiple segmentation definitions (e.g. ±5 units of normalized/diplomatic text or ±5 units of speaker1/speaker2/… etc.
  • Automatically or manually generated example queries per corpus on welcome screen
  • List of available metadata added to Corpus Explorer
  • Support for “corpus collections” (thematic groups of corpora in corpora list)
  • Support for server-side embedded fonts
  • Visualizations can be hidden/shown by default, option to use grid as default view
  
  
• Query language and query interface:
  • Typed precedence operators (e.g. #1 .sentence,1,2 #2)
  • Support for user defined and randomly generated example queries
  • Virtual keyboard
  • Visible query links (copy & paste link from browser)
  • Matching document count for search results (e.g. "105 hits in 22 documents")