Mining Semantic Structures from Syntactic Structures in Free Text Documents

Abstract

Abstract-The Web has made possible many advanced textmining applications, such as news summarization, essay grading, question answering, and semantic search. For many of such applications, statistical text-mining techniques are ineffective since they do not utilize the morphological structure of the text. Thus, many approaches use NLP-based techniques, that parse the text and use patterns to mine and analyze the parse trees which are often unnecessarily complex. Therefore, we propose a weighted-graph representation of text, called TextGraphs, which captures the grammatical and semantic relations between words and terms in the text. TextGraphs are generated using a new text mining framework which is the main focus of this paper. Our framework, SemScape, uses a statistical parser to generate few of the most probable parse trees for each sentence and employs a novel two-step pattern-based technique to extract from parse trees candidate terms and their grammatical relations. Moreover, SemScape resolves coreferences by a novel technique, generates domain-specific TextGraphs by consulting ontologies, and provides a SPARQL-like query language and an optimized engine for semantically querying and mining TextGraphs.

Harvesting Domain Specific Ontologies from Text

“…By integrating these links, SemScape generates the TextGraphs such as the one depicted in Figure 1. Task 4: SemScape also provides techniques for Anaphora and Co-reference Resolution to improve the TextGraphs by replacing pronouns and coreferences with the terms they are referring to [23] [24]. E.g., the pronoun "that" in the running example is resolved with "expression".…”

“…SemScape is also able to generate general semantic links between the terms in the TGs [21] [24]. The most prominent example of such semantic links is the <subject, verb, object> link, in which the verb in a sentence is connecting its subjects to its objects.…”

“…The main tasks performed in each iteration by OntoHarverster can be summarized as follows: 1) Building TextGraphs: This task is performed by using our text-mining system SemScape [22] [24]. SemScape uses a pattern-based technique, to capture candidate terms and their grammatical links in the text, and represents them as weighted hyper-graphs, called TextGraphs.…”

Text-Mining, Structured Queries, and Knowledge Management on Web Document Corpora

“…This step is performed using SemScape, which uses Co-reference Resolution and other novel techniques to generate high-quality TextGraphs [22]. TextGraphs generated in this way provide a semantic representation of the grammatical connections between words, terms, and phrases through labeled and weighted links.…”

“…To improve coverage of our integrated KB, we developed the IBminer and the OntoMiner systems, which generate InfoBoxes and ontologies from free text. IBminer employs an NLP-based text mining engine called SemScape [22] to identify the morphological information in text and generate graph-based structures called TextGraphs. Then, IBminer extracts semantic links from TextGraphs using predefined patterns and converts semantic links to final InfoBoxes.…”