Compare the Structure of Sentences
The structure of two sentences can be compared by computing and processing their respective constituent graphs.
Display the constituent tree of a sentence as a graph.
graph = TextStructure["Time flies like an arrow.", "ConstituentGraph"]
Compute the matrix of distances between all the vertices of this graph.
distancemat1 = GraphDistanceMatrix[First[graph]]; MatrixForm[distancemat1]
Proceed similarly for another sentence.
graph2 = TextStructure["I fly in the sky.", "ConstituentGraph"]; distancemat2 = GraphDistanceMatrix[First[graph2]];
Compare the structure of the two sentences by comparing their distance matrices.
distancemat1 == distancemat2
The two sentences have the same structure.
Find sentences with identical structures in two different Wikipedia articles. First, extract sentences with a given word count and generate the constituent graph for each of them.
processWikiPage[article_] := Select[TextCases[WikipediaData[article], "Sentences"], WordCount[#] < 5 &]; genStructure[article_] := Flatten[TextStructure[#, "ConstituentGraph"] & /@ processWikiPage[article]];
phrasestruct1 = genStructure["Philosophy"]; phrasestruct2 = genStructure["History"];
Compute all distance matrices.
adj1 = GraphDistanceMatrix /@ phrasestruct1; adj2 = GraphDistanceMatrix /@ phrasestruct2;
Compare the sentences of the different articles two by two.
comparison = Outer[Equal, adj1, adj2, 1];
These are the pairs of sentences with the same structure.
pickedSentences = Flatten[Pick[Outer[List, phrasestruct1, phrasestruct2], comparison, True], 1];
This is the first pair.