{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/88091"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/88091","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Expanding commonsense knowledge bases by learning from image tags","abstract":"I present a method for learning new commonsense facts to augment existing commonsense knowledge bases by using the metadata of large online image collections. Online image collections present a source of knowledge that is supported by many contributors, has good representation of objects and their properties, and is visual. The collection's broad support of objects and object properties ensure the relevance and quality of the commonsense knowledge collected, while the visual focus provides a different subset of knowledge than typical text corpora. Using the image metadata provides a text representation of the visual information. Therefore, I can use classifiers trained on existing text-based knowledge bases to learn relationships between concepts represented in the images. I collect two datasets of more than 1 million images each, one consisting of animal images, one of room interiors. The images are tagged with relevant concepts by their owners. I train classifiers using facts from two popular commonsense knowledge bases, ConceptNet and Freebase, to classify the relationships between frequent concept pairs. The output is a list of more than 90,000 proposed facts, which are in neither source knowledge base.","abstract_html":"I present a method for learning new commonsense facts to augment existing commonsense knowledge bases by using the metadata of large online image collections. Online image collections present a source of knowledge that is supported by many contributors, has good representation of objects and their properties, and is visual. The collection's broad support of objects and object properties ensure the relevance and quality of the commonsense knowledge collected, while the visual focus provides a different subset of knowledge than typical text corpora. Using the image metadata provides a text representation of the visual information. Therefore, I can use classifiers trained on existing text-based knowledge bases to learn relationships between concepts represented in the images. I collect two datasets of more than 1 million images each, one consisting of animal images, one of room interiors. The images are tagged with relevant concepts by their owners. I train classifiers using facts from two popular commonsense knowledge bases, ConceptNet and Freebase, to classify the relationships between frequent concept pairs. The output is a list of more than 90,000 proposed facts, which are in neither source knowledge base.","abstract_has_math":false,"creators":["Mauceri, Cecilia R"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":["Lazebnik, Svetlana"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2015,"date_issued":"2015-09-29T20:38:44Z","date_published":"2015-09-29T20:38:44Z","updated_at":"2026-07-22T22:26:31Z","subjects":["Commonsense knowledge","transfer learning","image collection","knowledge extraction"],"languages":["en"],"rights":["Copyright 2015 Cecilia Mauceri"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/2142/88091","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Lazebnik, Svetlana"]},{"key":"dc:creator","label":"Author","values":["Mauceri, Cecilia R"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2015-09-29T20:38:44Z","2015-08","2015-07-20","2015-8"]},{"key":"dc:type","label":"Dc Type","values":["text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Commonsense knowledge","transfer learning","image collection","knowledge extraction"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2015 Cecilia Mauceri"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/2142/88091"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["I present a method for learning new commonsense facts to augment existing commonsense knowledge bases by using the metadata of large online image collections. Online image collections present a source of knowledge that is supported by many contributors, has good representation of objects and their properties, and is visual. The collection's broad support of objects and object properties ensure the relevance and quality of the commonsense knowledge collected, while the visual focus provides a different subset of knowledge than typical text corpora. Using the image metadata provides a text representation of the visual information. Therefore, I can use classifiers trained on existing text-based knowledge bases to learn relationships between concepts represented in the images. I collect two datasets of more than 1 million images each, one consisting of animal images, one of room interiors. The images are tagged with relevant concepts by their owners. I train classifiers using facts from two popular commonsense knowledge bases, ConceptNet and Freebase, to classify the relationships between frequent concept pairs. The output is a list of more than 90,000 proposed facts, which are in neither source knowledge base.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2015-09-29 without embargo terms","The student, Cecilia Mauceri, accepted the attached license on 2015-07-20 at 14:10.","The student, Cecilia Mauceri, submitted this Thesis for approval on 2015-07-20 at 14:31.","This Thesis was approved for publication on 2015-07-20 at 14:51.","DSpace SAF Submission Ingestion Package generated from Vireo submission #8567 on 2015-09-29 at 13:23:18","Made available in DSpace on 2015-09-29T20:38:44Z (GMT). No. of bitstreams: 2 MAUCERI-THESIS-2015.pdf: 31250800 bytes, checksum: 31e742418985cc88b2f376b0b96bdab4 (MD5) LICENSE.txt: 4212 bytes, checksum: 69170c49fb8c2f18fffb69a27eec6e1c (MD5) Previous issue date: 2015-07-20"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Expanding commonsense knowledge bases by learning from image tags"]}]}],"canonical_facts":{"dc:contributor":["Lazebnik, Svetlana"],"dc:creator":["Mauceri, Cecilia R"],"dc:date":["2015-09-29T20:38:44Z","2015-08","2015-07-20","2015-8"],"dc:description":["I present a method for learning new commonsense facts to augment existing commonsense knowledge bases by using the metadata of large online image collections. Online image collections present a source of knowledge that is supported by many contributors, has good representation of objects and their properties, and is visual. The collection's broad support of objects and object properties ensure the relevance and quality of the commonsense knowledge collected, while the visual focus provides a different subset of knowledge than typical text corpora. Using the image metadata provides a text representation of the visual information. Therefore, I can use classifiers trained on existing text-based knowledge bases to learn relationships between concepts represented in the images. I collect two datasets of more than 1 million images each, one consisting of animal images, one of room interiors. The images are tagged with relevant concepts by their owners. I train classifiers using facts from two popular commonsense knowledge bases, ConceptNet and Freebase, to classify the relationships between frequent concept pairs. The output is a list of more than 90,000 proposed facts, which are in neither source knowledge base.","Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2015-09-29 without embargo terms","The student, Cecilia Mauceri, accepted the attached license on 2015-07-20 at 14:10.","The student, Cecilia Mauceri, submitted this Thesis for approval on 2015-07-20 at 14:31.","This Thesis was approved for publication on 2015-07-20 at 14:51.","DSpace SAF Submission Ingestion Package generated from Vireo submission #8567 on 2015-09-29 at 13:23:18","Made available in DSpace on 2015-09-29T20:38:44Z (GMT). No. of bitstreams: 2 MAUCERI-THESIS-2015.pdf: 31250800 bytes, checksum: 31e742418985cc88b2f376b0b96bdab4 (MD5) LICENSE.txt: 4212 bytes, checksum: 69170c49fb8c2f18fffb69a27eec6e1c (MD5) Previous issue date: 2015-07-20"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/2142/88091"],"dc:language":["en"],"dc:rights":["Copyright 2015 Cecilia Mauceri"],"dc:subject":["Commonsense knowledge","transfer learning","image collection","knowledge extraction"],"dc:title":["Expanding commonsense knowledge bases by learning from image tags"],"dc:type":["text"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:26:31Z"}