Description

Return to Download

Dependency Viewer is developed to visualize the dependency parsing result in CONLL format. It can be used to：
  • Visualize dependency sentence in horizontal view or tree view
  • Edit the dependency file
  • Export the view to image in PNG/XPS and even EMF(vector image) format
  • Show two results at the same time for comparation
  • Advanced search function
  • Statistic and evaluation

    • The tool is very easy to use while it's a little hard to use the advanced search function. The following will describe the search function in detail.

    Definition of Search

    The search function can be described in the following grammar.

    Matcher=CountMatcher OR_Matcher

    CountMatcher=SentenceLengthMatcher:LabeledErrorCountMatcher:UnlabeledErrorCountMatcher//to match sentence length and error count, for example, {5,12}:{,8}:{2} is used to match sentences with length between 5 to 12, labeled error count smaller or equal to 8 and unlabeled error count exactly be 2

    SentenceLengthMatcher={MinValue,MaxValue}

    SentenceLengthMatcher={,MaxValue}

    SentenceLengthMatcher={MinValue,}

    SentenceLengthMatcher={ExactValue}

    SentenceLengthMatcher=NULL

    LabeledErrorCountMatcher=SentenceLengthMatcher

    UnlabeledErrorCountMatcher=SentenceLengthMatcher

    OR_Matcher=OR_Matcher || OR_Matcher//The query does not support multi-level combination of AND/OR. The first level is always OR and the second is always AND. The combination between CountMatcher and OR_Matcher is AND.

    OR_Matcher=NULL

    OR_Matcher=AND_Matcher

    AND_Matcher=AND_Matcher && AND_Matcher

    AND_Matcher=NULL

    AND_Matcher=SequenceMatcher

    SequenceMatcher=DependencyMatcher SequenceMatcher//SequenceMatcher is used to search sentence structure. For example, 政策->变更 行政院(.2,) will find sentences with "政策" which is dependent to "变更", and "行政院" two after "政策". Another example is 立法院 听取 报告 which means "报告" one after "听取" and "听取" one after "立法院"(when there is no PositionMatcher, one after relation is by default)

    SequenceMatcher=NULL

    DependencyMatcher=WordMatcher->DependencyMatcher

    DependencyMatcher=NULL

    WordMatcher=WordFormMatcherWordPosTagMatcherPositionMatcherDependencyLabelMatcher

    WordMatcher=WordFormMatcherWordPosTagMatcherPositionMatcherDependencyLabelMatcher#WordFormMatcherWordPosTagMatcherPositionMatcherDependencyLabelMatcher
    //The content after # indicates the WordMatcher in contrast sentence. It can be used to match errors and differences. For example, 可->谓#不 indicates that "可" is dependent to "谓" while it's dependent to "不" in contrast sentence. 赵宁<d-genetive>#<d-agent> means that the dependency label of "赵宁" is "d-genetive" while in contrast sentence it's "d-agent"

    WordFormMatcher=Regex Matcher for Word's Form

    WordFormMatcher=NULL

    WordPosTagMatcher=[Regex Matcher for Word's PosTag]

    WordPosTagMatcher=NULL

    DependencyLabelMatcher=[Regex Matcher for Word's Dependency Label]

    DependencyLabelMatcher=NULL

    PositionMatcher=(Min Position,Max Position)

    PositionMatcher=(,Max Position)

    PositionMatcher=(Min Position,)

    PositionMatcher=(Exactly Position)

    PositionMatcher=NULL

    Position=AbsolutePosition

    Position=RelativePosition

    AbsolutePosition=number//AbsolutePosition is used to describe the index of word in sentence starting from 1. Negtive number means start from the end.

    RelativePosition=.number//RelativePosition is used to describe the index of a word related to the previous word. .1 indicates the next one while .-1 indicates the previous one.

    //"Search In Contrast Dependency File" just swaps the depedency file and contrast dependency file.


    Examples of Search

    Search: "投资" one after "外商" ("外商 投资")
    Search In Dependency File: 外商 投资

    Search: sentence contains "外商" and "青海"
    Search In Dependency File: 外商 && 青海

    Search: sentence contains "外商" or "青海"
    Search In Dependency File: 外商 || 青海

    Search: sentence contains "外商" or "青海" with length between 2 to 8
    Search In Dependency File: {2,8}:: 外商 || 青海

    Search: sentences whoes length is equal to or longer than 5
    Search In Dependency File: {5,}::

    Search: sentences whoes length is 5
    Search In Dependency File: {5}::

    Search: sentences whoes length is equal to or smaller than 10
    Search In Dependency File: {,10}::

    Search: contains "外商" and its PosTag is "NN"
    Search In Dependency File: 外商[NN]

    Search: contains "外商" and its dependency label is "agent"
    Search In Dependency File: 外商<agent>

    Search: contains "外商" and its position in sentence is between 2 and 8
    Search In Dependency File: 外商(2,8)

    Search: "外商" is dependent to "投资" and "投资" is dependent to "热点"
    Search In Dependency File: 外商->投资->热点

    Search: "外商" is dependent to "投资" whoes PosTag is "VV" and "投资" is dependent to "热点" which is three after "投资"
    Search In Dependency File: 外商->投资[VV]->热点(.3)

    Search: sentences whoes length is between 2 to 10 with labeled error count from 3 to 8 and unlabeled error count from 2 to 4
    Search In Dependency File: {2,10}:{3,8}:{2,4}

    Search: "外商" is dependent to "投资" with dependency label "agent" and "投资" is dependent to "热点". In the contrast sentence, "外商" is dependent to "成为" with label "gagent", and "投资" is dependent to "新"
    Search In Dependency File: 外商<agent>#<gagent>->投资#成为->热点#新

    Search: In the contrast sentence, "外商" is dependent to "投资" with dependency label "agent" and "投资" is dependent to "热点". In the original dependency sentence, "外商" is dependent to "成为" with label "gagent", and "投资" is dependent to "新"
    Search In Contrast Dependency File: 外商<agent>#<gagent>->投资#成为->热点#新