Package edu.upf.taln.dri.lib.model.graph

Class DependencyGraph

  • public class DependencyGraph
extends Object
    Dependency Graph data structure and methods

    • Constructor Detail

      • DependencyGraph

        public DependencyGraph()

    • Method Detail

      • getNodeName

        public String getNodeName​(Integer nodeId)
        Get node name
        Parameters:
        nodeId -
        Returns:

      • changeNodeName

        public boolean changeNodeName​(Integer nodeId,
                              String newNodeName)
        Change the name of the node - new name not null or empty
        Parameters:
        nodeId -
        newNodeName -
        Returns:

      • setEdgeSRLsenseAndTag

        public Boolean setEdgeSRLsenseAndTag​(Integer edgeId,
                                     String tag,
                                     String sense,
                                     Integer sentenceId)
        Set an SRL edge sense and tag and source sentence ID
        Parameters:
        edgeId -
        tag -
        sense -
        sentenceId -
        Returns:

      • setNodeInCrossSentCausalRel

        public Boolean setNodeInCrossSentCausalRel​(Integer nodeId,
                                           Integer causalRelId,
                                           String roleName,
                                           Integer causeNode,
                                           Integer effectNode)
        Set a node as participating to a cross-sentence causal relation (the cause annotated in a sentence and the effect in another one) A node annotated with these properties can be the cause or the effect of the causal relation.
        Parameters:
        nodeId - the ID of the node that participates in the causal relation
        causalRelId - the ID of the causal relation
        roleName - the name of the role of the node in the causal relation
        causeNode - the id of the node that identifies the cause of the causal relation
        effectNode - the id of the node that identifies the effect in the causal relation
        Returns:

      • addRhetoricalClass

        public Boolean addRhetoricalClass​(Integer nodeId,
                                  String rhetoricalClass,
                                  boolean deletePreexisting)
        Add a rhetorical class to the set of rhetorical classes associated to the node
        Parameters:
        nodeId -
        rhetoricalClass - if null or empty, no rhetorical class is added (to empty the rhetorical class set, set to null and deletePreexisting to true)
        deletePreexisting - if true, all pre-existing rhetorical classes are deleted
        Returns:

      • getRhetoricalClassSet

        public Set<String> getRhetoricalClassSet​(Integer nodeId)
        Get the set of rhetorical classes associated to the node
        Parameters:
        nodeId -
        Returns:

      • addHeadWord

        public Boolean addHeadWord​(Integer nodeId,
                           String headWord,
                           boolean deletePreexisting)
        Add a head word to the set of head words associated to the node. Usually, each node not derived from the merging of coreferents has only a head word.
        Parameters:
        nodeId -
        headWord - if null or empty, no head word is added (to empty the head words set, set to null and deletePreexisting to true)
        deletePreexisting - if true, all pre-existing head words are deleted
        Returns:

      • getHeadWordsSet

        public Set<String> getHeadWordsSet​(Integer nodeId)
        Get the set of head words associated to the node
        Parameters:
        nodeId -
        Returns:

      • setSentenceIDTokensPair

        public Boolean setSentenceIDTokensPair​(Integer nodeId,
                                       Integer sentenceID,
                                       List<String> sentenceTokenList)
        Set the pair of sentence ID and list of Tokens for a specific node (list of sentence words separated by space)
        Parameters:
        nodeId -
        sentenceID -
        sentenceTokenList -
        Returns:

      • getSentenceIDTokensPair

        public org.apache.commons.lang3.tuple.Pair<Integer,​List<String>> getSentenceIDTokensPair​(Integer nodeId)
        Get the pair of sentence ID and list of Tokens of a specific node (list of sentence words separated by space)
        Parameters:
        nodeId -
        Returns:

      • addEdge

        public Integer addEdge​(String name,
                       Integer from,
                       Integer to,
                       boolean isSRL)
        Add edge
        Parameters:
        name -
        from -
        to -
        isSRL -
        Returns:

      • getEdgeName

        public String getEdgeName​(Integer edgeId)
        Get edge name
        Parameters:
        edgeId -
        Returns:

      • changeEdgeName

        public boolean changeEdgeName​(Integer edgeId,
                              String newEdgeName)
        Change the name of the edge - new name not null or empty
        Parameters:
        edgeId -
        newEdgeName -
        Returns:

      • getAllOrderedDepthFirstEdges

        public List<Integer> getAllOrderedDepthFirstEdges()
        Get all edges of the dependency graph ordered by depth first visit
        Parameters:
        edgeId - ordered list of edge ids
        Returns:

      • getAllEdges

        public Map<Integer,​org.apache.commons.lang3.tuple.Pair<Integer,​Integer>> getAllEdges()
        Get all edges of the dependency graph
        Parameters:
        edgeId -
        Returns:

      • getEdgeFromNode

        public Integer getEdgeFromNode​(Integer edgeId)
        Get the start node of the edge
        Parameters:
        edgeId -
        Returns:

      • getEdgeToNode

        public Integer getEdgeToNode​(Integer edgeId)
        Get the end node of the edge
        Parameters:
        edgeId -
        Returns:

      • getEdgeSRLtag

        public String getEdgeSRLtag​(Integer edgeId)
        Get the SRL tag of the edge
        Parameters:
        edgeId -
        Returns:

      • getEdgeSRLsensefeat

        public String getEdgeSRLsensefeat​(Integer edgeId)
        Get the SRL sense of the edge
        Parameters:
        edgeId -
        Returns:

      • getEdgeSRLsentId

        public Integer getEdgeSRLsentId​(Integer edgeId)
        Get the SRL sent ID of the edge
        Parameters:
        edgeId -
        Returns:

      • getNodePOS

        public String getNodePOS​(Integer nodeId)
        Get node POS
        Parameters:
        nodeId -
        Returns:

      • getNodeLemma

        public String getNodeLemma​(Integer nodeId)
        Get node POS
        Parameters:
        nodeId -
        Returns:

      • getNodeCorefID

        public Set<Integer> getNodeCorefID​(Integer nodeId)
        The IDs of the coreference chains in which the node is the head of one of their elements. Usually it should be a one element List since each node can be head of one element of a specific coreference chain.
        Parameters:
        nodeId -
        Returns:

      • getNodeCorefName

        public String getNodeCorefName​(Integer nodeId)
        Get co-reference node name (aggregates all the tokens of the coreference chain element).
        Parameters:
        nodeId -
        Returns:

      • getNodeSentOrder

        public Integer getNodeSentOrder​(Integer nodeId)
        Get node sentence order
        Parameters:
        nodeId -
        Returns:

      • getNodeCount

        public int getNodeCount()
        Number of nodes in the graph
        Returns:

      • getRootNodeIds

        public Set<Integer> getRootNodeIds()
        Get IDs of root nodes
        Returns:

      • getEdgesByNameSourceAndDestination

        public Set<Integer> getEdgesByNameSourceAndDestination​(Integer sourceId,
                                                       Integer destinationId,
                                                       String name)
        Get edges by source node id / destination node id / name At least on of source node Id / destination node Id / name has not to be blank / empty. (AND OF source node id, destination node id and name checks)
        Parameters:
        edgeList - list of edges to filter
        sourceId -
        destinationId -
        name -
        Returns:
        filtered edge ids list

      • getEdgesByNameRegExp

        public Set<Integer> getEdgesByNameRegExp​(String nameRegExp)
        Get the id of the edges by regexp match on edge name
        Parameters:
        nameRegExp -
        Returns:

      • getNodesByNameRegExp

        public Set<Integer> getNodesByNameRegExp​(String nameRegExp)
        Get the id of the nodes by regexp match on node name
        Parameters:
        nameRegExp -
        Returns:

      • getParentNodeIds

        public Set<Integer> getParentNodeIds​(Integer nodeId)
        The id(s) of the parent node(s)
        Parameters:
        nodeId -
        Returns:

      • getChildrenNodes

        public Set<Integer> getChildrenNodes​(Integer nodeId)
        The id(s) of the children node(s)
        Parameters:
        nodeId -
        Returns:

      • getIncidentEdges

        public Map<Integer,​org.apache.commons.lang3.tuple.Pair<Integer,​Integer>> getIncidentEdges​(Integer nodeId,
                                                                                                      String name)
        Get the set of edges entering in the node. If name is not null or empty, only edges with that name are returned
        Parameters:
        nodeId -
        name -
        Returns:

      • getOutgoingEdges

        public Map<Integer,​org.apache.commons.lang3.tuple.Pair<Integer,​Integer>> getOutgoingEdges​(Integer nodeId,
                                                                                                      String name)
        Get the set of edges going out from the node. If name is not null or empty, only edges with that name are returned
        Parameters:
        nodeId -
        name -
        Returns:

      • getSRLframeRoots

        public Map<Integer,​Set<String>> getSRLframeRoots()
        Get a map of node ID (key) / set of sense by considering only nodes that are roots of SRL frames The set of senses is the set of different frame the node participates in as root.
        Returns:

      • getSRLframeParticipantTags

        public Map<Integer,​Set<Integer>> getSRLframeParticipantTags​(Integer rootNodeId)
        Get a map of predicates the node specified by ID is root of. Each element of the map has as key an unambiguous id of the SRL frame (ID of the sentence that contains the SRL frame) and as value the set of ids of the edges participating in the frame
        Parameters:
        rootNodeId -
        Returns:

      • graphAsString

        public String graphAsString​(GraphToStringENUM outputType)
        Get a string representation of the graph
        Parameters:
        formatType -
        Returns:

      • deleteEdgesByNameRegExp

        public int deleteEdgesByNameRegExp​(List<String> regexpList,
                                   boolean deleteMatching)
        Delete all the edges with name that matches one of the reg exps
        Parameters:
        regexpList -
        deleteMatching - if false delete the edges that doesn't match
        Returns:

      • mergeNodes

        public boolean mergeNodes​(Integer nodeId1,
                          Integer nodeId2,
                          String newNodeName)
        Merge the second node with the first. The properties of the first node are preserved. In-going and outgoing edges of the second node are moved to the first one. NB: the new name of a merged pair of nodes is consistent only if both nodes are in the same sentence. Since the sentence specific order is exploited to determine the order of the names of the merged nodes.
        Parameters:
        nodeId1 -
        nodeId2 -
        newNodeName - if not null, the node 1 will be renamed
        Returns:

      • getMergedIDmap

        public Map<Integer,​String> getMergedIDmap​(Integer nodeId)
        Get the map of original nodeID and lexicalization of the nodes merged in the current node
        Parameters:
        nodeId -
        Returns:

      • addToMergedNameMap

        public boolean addToMergedNameMap​(Integer nodeId,
                                  Integer mergedNodeId,
                                  String mergedNodeName,
                                  boolean emptyMap)
        Add an element to the node_mergedNameMap of the node
        Parameters:
        nodeId -
        nodeName -
        emptyMap -
        Returns:

      • addToMergedIDmap

        public boolean addToMergedIDmap​(Integer nodeId,
                                Integer mergedNodeId,
                                String mergedNodeName,
                                boolean emptyMap)
        Add an element to the node_mergedIDmap of the node
        Parameters:
        nodeId -
        nodeName -
        emptyMap -
        Returns:

      • getMergedNameMap

        public Map<Integer,​String> getMergedNameMap​(Integer nodeId)
        Get the map of sentence order ID of node and lexicalization of the sentence nodes merged in the current node
        Parameters:
        nodeId -
        Returns:

      • getCausalRoleNameMap

        public Map<Integer,​String> getCausalRoleNameMap​(Integer nodeId)
        Get the map of causal relations the nodes is part of. The keys are the ID of each causal relation, while the values are the role of the node in each causal relation (CAUSE or EFFECT).
        Parameters:
        nodeId -
        Returns:

      • getCausalCauseIDmap

        public Map<Integer,​Integer> getCausalCauseIDmap​(Integer nodeId)
        Get the map of causal relations the nodes is part of. The keys are the ID of each causal relation, while the values are the ID of the cause node in each causal relation.
        Parameters:
        nodeId -
        Returns:

      • getCausalEffectIDmap

        public Map<Integer,​Integer> getCausalEffectIDmap​(Integer nodeId)
        Get the map of causal relations the nodes is part of. The keys are the ID of each causal relation, while the values are the ID of the cause node in each causal relation.
        Parameters:
        nodeId -
        Returns:

      • sanitizeGraph

        public void sanitizeGraph()
        Check and remove self loops edges and merges duplicated / non-SRL edges

      • compactNodes

        public Integer compactNodes()
        Tentative version of node collapsing heuristics over a dependency graph
        Returns:

      • main

        public static void main​(String[] args)