PrOntoLearn: Unsupervised Lexico-Semantic Ontology Generation Using Probabilistic Methods

Abstract

Abstract.Formalizing an ontology for a domain manually is well-known as a tedious and cumbersome process. It is constrained by the knowledge acquisition bottleneck. Therefore, researchers developed algorithms and systems that can help to automatize the process. Among them are systems that include text corpora for the acquisition. Our idea is also based on vast amount of text corpora. Here, we provide a novel unsupervised bottom-up ontology generation method. It is based on lexico-semantic structures and Bayesian reasoning to expedite the ontology generation process. We provide a quantitative and two qualitative results illustrating our approach using a high throughput screening assay corpus and two custom text corpora. This process could also provide evidence for domain experts to build ontologies based on top-down approaches.

Concept Meaning Acquisition based on HowNet and Its Application in the Construction of Taxonomy

“…However, the ability of transplanting to other fields is poor, and the recall rate may decline greatly. The most frequently used methods for construct taxonomy include top-down [3] or bottom-up [1] hierarchical clustering algorithms [21] by calculating the similarity of concepts. It still has problems of sparse data and tightness between classes is difficult [17].…”

RLLib: C++ Library to Predict, Control, and Represent Learnable Knowledge Using On/Off Policy Reinforcement Learning

“…Therefore, the memory footprint as well as the computational requirements that are needed by the library have been optimized; (2) a configurable C++ template functions to synchronize with application or device hardware requirements; (3) the library emphasizes more on learnable knowledge representation and reasoning from sensorimotor streams; (4) a clean and transparent API exists that enables users to model their RL problems easily; and (5) a self-contained C++ template library covers plethora of incremental, standard, and gradient temporal-difference learning algorithms in RL that is published to-date (e.g., [22,28]). Our library has been successfully used in [1] to learn role assignment in RoboCup 3D soccer simulation agents.…”

Formal Ontology Learning from English IS-A Sentences

“…Comprehensive and consistent domain-specific ontologies are useful in building applications that can perform complex tasks such as question answering [1], machine translation [2], bio-medical knowledge mining [3], and other Semantic Web applications. However, manually creating ontologies that may comprise of a large number of concepts and relations is a highly labor intensive and expensive task.…”

“…Some of the linguistic non-trivialities that are commonly observed are: (i) lexical variations of quantifiers and "is a" lexemes, (ii) lexico-syntactic variations of IS-A patterns such as some of the Hearst patterns [8], (iii) tense variations of "is a" lexemes such as "was", "had become", (iv) modal variations of "is a" lexemes such as "may be", "could be", and (v) comparative and superlative constructs of IS-A sentences. We demonstrate that OL on IS-A sentences provides the axiomatic foundation for formally representing factual non IS-A sentences 3 as well, and therefore requires special study. To support this argument, we created a community IS-A test dataset and ran contemporary state-of-the-art general-purpose OL tools, but with unsatisfactory results with respect to standard accuracy measures.…”