From Unstructured Web Knowledge to Plan Descriptions

Abstract

Automated Planning (AP) is an AI field whose goal is to automatically generate sequence of actions that solve problems. One of the main difficulties in its extensive use in real-world application lies in the fact that it requires the careful and error-prone process of defining a declarative domain model. This is usually performed by planning experts who should know about both the domain in hand, and the planning techniques (including sometimes the inners of these techniques or the tools that implement them). In order planning to be widely used, this process should be performed by non-planning experts. On the other hand, in many domains there are plenty of electronic documents (including the Web) that describe processes or plans in a semi-structured way. These descriptions mix natural language and certain templates for that specific domain. One such examples is the www.WikiHow.com web site that includes plans in many domains, all plans described through a set of common templates. In this work, we present a suite of tools that automatically extract knowledge from those unstructured descriptions of plans to be used for diverse planning applications.

Integrating Know-How into the Linked Data Cloud

“…Several research projects have already attempted to extract human know-how from manually generated instructions. One of the most frequent approaches, based on Natural Language Processing (NLP), has been used to extract knowledge from domain-independent know-how repositories, such as wikiHow 5 [1], [6] as well as domain-specific ones, like the medical domain [12]. An approach based on statistical analysis has also been used to extract procedural knowledge from a more diverse set of Web documents not necessarily focused on know-how [3].…”

A semantic web of know-how

“…Unstructured know-how is often represented in natural language. Most of the existing approaches for extracting procedural knowledge from natural language texts use Natural Language Processing (NLP) techniques [1,5,9]. Other approaches have also combined NLP techniques with Machine Learning [10] and statistical analysis [2].…”

“…This paper addresses this problem proposing a novel knowledge representation framework that can effectively represent community know-how. The main goals of this paper are (1) to motivate the need for a semantic representation of know-how, (2) to describe the features of the proposed framework and (3) to demonstrate its feasibility.…”

Vocabulary Alignment for Collaborative Agents: a Study with Real-World Multilingual How-to Instructions

“…This scenario is based on previous research which showed how step-by-step instructions described in natural language can be automatically formalised into machine understandable data [Pareti et al, 2014]. Unlike related work in processing instructional knowledge [Addis and Borrajo, 2011;Kiddon et al, 2015;Tenorth et al, 2010;Schumacher et al, 2012;Malmaud et al, 2014], the work in [Pareti et al, 2014] provides a formalisation that allows agents to automatically understand and execute instructions, provided they have the necessary abilities [Pareti et al, 2016;Pareti, 2016]. We consider a publicly available dataset that follows this formalisation and that has been extracted from instructional websites.…”