Never-ending learning

Abstract

Whereas people learn many different types of knowledge from diverse experiences over many years, and become better learners over time, most current machine learning systems are much more narrow, learning just a single function or data model based on statistical analysis of a single data set. We suggest that people learn better than computers precisely because of this difference, and we suggest a key direction for machine learning research is to develop software architectures that enable intelligent agents to also learn many types of knowledge, continuously over many years, and to become better learners over time. In this paper we define more precisely this never-ending learning paradigm for machine learning, and we present one case study: the Never-Ending Language Learner (NELL), which achieves a number of the desired properties of a never-ending learner. NELL has been learning to read the Web 24hrs/ day since January 2010, and so far has acquired a knowledge base with 120mn diverse, confidence-weighted beliefs (e.g., servedWith(tea,biscuits)), while learning thousands of interrelated functions that continually improve its reading competence over time. NELL has also learned to reason over its knowledge base to infer new beliefs it has not yet read from those it has, and NELL is inventing new relational predicates to extend the ontology it uses to represent beliefs. We describe the design of NELL, experimental results illustrating its behavior, and discuss both its successes and shortcomings as a case study in never-ending learning. NELL can be tracked online at http://rtw.ml.cmu.edu, and followed on Twitter at @CMUNELL. 2. RELATED WORK Previous research has considered the problem of designing machine learning agents that persist over long periods research highlights

From Textbooks to Knowledge: A Case Study in Harvesting Axiomatic Knowledge from Textbooks to Solve Geometry Problems

“…Information Extraction from Textbooks: Our model builds upon ideas from Information extraction (IE), which is the task of automatically extracting structured information from unstructured and/or semi-structured documents. While there has been a lot of work in IE on domains such as web documents (Chang et al, 2003;Etzioni et al, 2004;Cafarella et al, 2005;Chang et al, 2006;Banko et al, 2007;Etzioni et al, 2008;Mitchell et al, 2015) and scientific publication data (Shah et al, 2003;Peng and McCallum, 2006;Saleem and Latif, 2012), work on IE from educational material is much more sparse. Most of the research in IE from educational material deals with extracting simple educational concepts (Shah et al, 2003;Canisius and Sporleder, 2007;Liu et al, 2016b;Wang et al, 2016) or binary relational tuples (Balasubramanian et al, 2002;Clark et al, 2012;Dalvi et al, 2016) using existing IE techniques.…”

Identifying Equivalent Relation Paths in Knowledge Graphs

“…Also, its semantics "a city has a river, and city has a generalization" is different from the original relation path query semantics. In NELL [19], the relation generalizations is common and acts as a meta relation for classes of entities, e.g., the generalization of a city is Location. So, the second part of P is a loop decoded as "a city is a location, and location is a generalization of a city".…”

“…A possible reason is that in KGs with low depth relation paths are rather short and more prone to contain loops with inverse relations. Here, we use four commonly used and general human knowledge KGs of different size and domain (Table 1 shows their characteristics): (A) Never Ending Learning (NELL) [19] is a knowledge base generated continuously by a never-ending machine learning system that crawls web pages and extracts facts from a pre-defined set of categories and relations; (B) YAGO3 [16] applies similar techniques to NELL, but it limits its sources to Wikipedia and WordNet [18]; (C) DBpedia [13] is a crowd-sourced knowledge base extracted from Wikipedia and Geonames as sources; and (D) WordNet [18] is a large lexical database of English, linking words and concepts using cognitive synonyms.…”

Merging knowledge bases in different languages

“…Among others, information extraction systems (IES) which use the text found in webpages to extract, validate and incorporate beliefs to a structured knowledge base have been developed (e.g., YAGO (Suchanek et al, 2008), NELL (Mitchell et al, 2015) or Knowledge Vault (Dong et al, 2014)). Such knowledge bases (KBs) store facts about the real world, which are represented as entities and relationship among entities.…”

“…Then, the proposed solutions are presented. In Section 4, the different approaches are tested in a case study with two KBs independently learnt by NELL (Mitchell et al, 2015) from English and Portuguese webpages respectively. Next, their behavior is discussed.…”