Commonsense Knowledge Base Research Articles

CogNet: Bridging Linguistic Knowledge, World Knowledge and Commonsense Knowledge

In this paper, we present CogNet, a knowledge base (KB) dedicated to integrating three types of knowledge: (1) linguistic knowledge from FrameNet, which schematically describes situations, objects and events. (2) world knowledge from YAGO, Freebase, DBpedia and Wikidata, which provides explicit knowledge about specific instances. (3) commonsense knowledge from ConceptNet, which describes implicit general facts. To model these different types of knowledge consistently, we introduce a three-level unified frame-styled representation architecture. To integrate free-form commonsense knowledge with other structured knowledge, we propose a strategy that combines automated labeling and crowdsourced annotation. At present, CogNet integrates 1,000+ semantic frames from linguistic KBs, 20,000,000+ frame instances from world KBs, as well as 90,000+ commonsense assertions from commonsense KBs. All these data can be easily queried and explored on our online platform, and free to download in RDF format for utilization under a CC-BY-SA 4.0 license. The demo and data are available at http://cognet.top/.

COVID19-Routes: A Safe Pedestrian Navigation Service

COVID-19 has become a global pandemic during 2020 due to its high contagiousness and the high mobility of the world's population today. In just one year, this virus has caused millions of infections and deaths worldwide. These numbers will continue to grow until the population becomes immune to the virus thanks to an effective vaccine. Until this is possible, the only viable strategy is to try to stop its expansion through preventive measures such as limiting mobility, the use of masks, etc. In order to support these measures, this article presents a service to provide safe navigation solutions to reduce the likelihood of infection by avoiding potential conflict areas in the city. To identify these hotspots, a strategy that combines a rule-based system and a common-sense knowledge base is proposed. Through this strategy, an occupation model and a danger model are inferred. This requires the prior capture of knowledge about the general functioning of the city, its inhabitants and the virus. The proposed service makes decisions from these two models. Finally, a validation process has been carried out through surveys to evaluate the proposed solution. Obtained results demonstrate the potential of the proposed solution as a tool to identify safe routes that allow citizens to move around the city with low exposure to COVID-19.

HealthAid: Extracting domain targeted high precision procedural knowledge from on-line communities

Recent advances in semantic web have shown how entity related searches have benefited from entity-based knowledge graphs. However, much of the commonsense knowledge about the real world is in the form of procedures or sequences of actions. Also, search log analysis shows that ‘how-to queries’ make up a significant amount of users’ queries. Unfortunately, these kinds of knowledge are missing from most knowledge graphs and commonsense knowledge bases in use. To empower semantic search, and other intelligent applications, computers need a much broader understanding of the world properties of everyday objects, human activities, and more. Luckily, such knowledge is abundantly available on-line and can be accessed from how-to communities. One domain of interest by on-line communities is the health domain, whereby users usually seek home remedies to common health-related issues. An example of such queries might be ‘how to stop nausea using acupressure’ or ‘how to aid digestion naturally’. To answer such questions, we need systems that understand natural language and knowledge bases with task frames of solutions in a holistic approach, including the tools required, the agents involved, and the temporal order of the actions. Our goal is to construct a machine-readable domain targeted high precision procedural knowledge base containing task frames. We developed a pipeline of methods leveraging open information extraction tool to extract procedural knowledge by tapping into on-line communities. Also, we devised a mechanism to canonicalize the task frames into clusters based on the similarity of the problems they intend to solve. The resulting know-how knowledge base, HealthAidKB, consists of more than 71 K task frames which are structured hierarchically and categorically; and can be used in many applications such as semantic search, digital personal assistants, human-computer dialog and computer vision. A comprehensive evaluation of our knowledge base shows high accuracy.

Understanding the Semantic Content of Sparse Word Embeddings Using a Commonsense Knowledge Base

Word embeddings have developed into a major NLP tool with broad applicability. Understanding the semantic content of word embeddings remains an important challenge for additional applications. One aspect of this issue is to explore the interpretability of word embeddings. Sparse word embeddings have been proposed as models with improved interpretability. Continuing this line of research, we investigate the extent to which human interpretable semantic concepts emerge along the bases of sparse word representations. In order to have a broad framework for evaluation, we consider three general approaches for constructing sparse word representations, which are then evaluated in multiple ways. We propose a novel methodology to evaluate the semantic content of word embeddings using a commonsense knowledge base, applied here to the sparse case. This methodology is illustrated by two techniques using the ConceptNet knowledge base. The first approach assigns a commonsense concept label to the individual dimensions of the embedding space. The second approach uses a metric, derived by spreading activation, to quantify the coherence of coordinates along the individual axes. We also provide results on the relationship between the two approaches. The results show, for example, that in the individual dimensions of sparse word embeddings, words having high coefficients are more semantically related in terms of path lengths in the knowledge base than the ones having zero coefficients.

Commonsense Knowledge Base Completion with Structural and Semantic Context

Automatic KB completion for commonsense knowledge graphs (e.g., ATOMIC and ConceptNet) poses unique challenges compared to the much studied conventional knowledge bases (e.g., Freebase). Commonsense knowledge graphs use free-form text to represent nodes, resulting in orders of magnitude more nodes compared to conventional KBs ( ∼18x more nodes in ATOMIC compared to Freebase (FB15K-237)). Importantly, this implies significantly sparser graph structures — a major challenge for existing KB completion methods that assume densely connected graphs over a relatively smaller set of nodes.In this paper, we present novel KB completion models that can address these challenges by exploiting the structural and semantic context of nodes. Specifically, we investigate two key ideas: (1) learning from local graph structure, using graph convolutional networks and automatic graph densification and (2) transfer learning from pre-trained language models to knowledge graphs for enhanced contextual representation of knowledge. We describe our method to incorporate information from both these sources in a joint model and provide the first empirical results for KB completion on ATOMIC and evaluation with ranking metrics on ConceptNet. Our results demonstrate the effectiveness of language model representations in boosting link prediction performance and the advantages of learning from local graph structure (+1.5 points in MRR for ConceptNet) when training on subgraphs for computational efficiency. Further analysis on model predictions shines light on the types of commonsense knowledge that language models capture well.

Constructing Uyghur Commonsense Knowledge Base by Knowledge Projection

Although considerable effort has been devoted to building commonsense knowledge bases (CKB), it is still not available for many low-resource languages such as Uyghur because of expensive construction cost. Focusing on this issue, we proposed a cross-lingual knowledge-projection method to construct an Uyghur CKB by projecting ConceptNet’s Chinese facts into Uyghur. We used a Chinese–Uyghur bilingual dictionary to get high-quality entity translation in facts and employed a back-translation method to eliminate the entity-translation ambiguity. Moreover, to tackle the inner relation ambiguity in translated facts, we made a hand-crafted rule to convert the structured facts into natural-language phrases and built the Chinese–Uyghur lingual phrases based on the similarity of phrases that corresponded to the bilingual semantic similarity scoring model. Experimental results show that the accuracy of our semantic similarity scoring model reached 94.75% for our task, and they successfully project 55,872 Chinese facts into Uyghur as well as obtain 67,375 Uyghur facts within a very short period.

The Role of Common-Sense Knowledge in Assessing Semantic Association

Natural language processing techniques often aim at automatically extracting semantics from texts. However, they usually need some available semantic knowledge contained in dictionaries and resources such as WordNet, Wikipedia, and FrameNet. In this respect, there is a large literature about the creation of novel semantic resources as well as attempts to integrate existing ones. In this context, we here focus on common-sense knowledge, which shows to have interesting characteristics as well as challenging issues such as ambiguity, vagueness, and inconsistency. In this paper, we make use of a large-scale and crowdsourced common-sense knowledge base, i.e., ConceptNet, to qualitatively evaluate its role in the perception of semantic association among words. We then propose an unsupervised method to disambiguate and integrate ConceptNet instances into WordNet, demonstrating how the enriched resource improves the recognition of semantic association. Finally, we describe a novel approach to label semantically associated words by exploiting the functional and behavioral information usually contained in common sense, demonstrating how this enhances the explanation (and the use) of relatedness and similarity with non-numeric information.

Classification of composite semantic relations by a distributional-relational model

Different semantic interpretation tasks such as text entailment and question answering require the classification of semantic relations between terms or entities within text. However, in most cases, it is not possible to assign a direct semantic relation between entities/terms. This paper proposes an approach for composite semantic relation classification using one or more relations between entities/term mentions, extending the traditional semantic relation classification task. The proposed model is different from existing approaches which typically use machine learning models built over lexical and distributional word vector features in that is uses a combination of a large commonsense knowledge base of binary relations, a distributional navigational algorithm and sequence classification to provide a solution for the composite semantic relation classification problem. The proposed approach outperformed existing baselines with regard to F1-score, Accuracy, Precision and Recall.

Augmenting End-to-End Dialogue Systems With Commonsense Knowledge

Building dialogue systems that can converse naturally with humans is a challenging yet intriguing problem of artificial intelligence. In open-domain human-computer conversation, where the conversational agent is expected to respond to human utterances in an interesting and engaging way, commonsense knowledge has to be integrated into the model effectively. In this paper, we investigate the impact of providing commonsense knowledge about the concepts covered in the dialogue. Our model represents the first attempt to integrating a large commonsense knowledge base into end-to-end conversational models. In the retrieval-based scenario, we propose a model to jointly take into account message content and related commonsense for selecting an appropriate response. Our experiments suggest that the knowledge-augmented models are superior to their knowledge-free counterparts.

Hierarchical Task Network planning with common-sense reasoning for multiple-people behaviour analysis

Safety on public transport is a major concern for the relevant authorities. We address this issue by proposing an automated surveillance platform which combines data from video, infrared and pressure sensors. Data homogenisation and integration is achieved by a distributed architecture based on communication middleware that resolves interconnection issues, thereby enabling data modelling. A common-sense knowledge base models and encodes knowledge about public-transport platforms and the actions and activities of passengers. Trajectory data from passengers is modelled as a time-series of human activities. Common-sense knowledge and rules are then applied to detect inconsistencies or errors in the data interpretation. Lastly, the rationality that characterises human behaviour is also captured here through a bottom-up Hierarchical Task Network planner that, along with common-sense, corrects misinterpretations to explain passenger behaviour. The system is validated using a simulated bus saloon scenario as a case-study. Eighteen video sequences were recorded with up to six passengers. Four metrics were used to evaluate performance. The system, with an accuracy greater than 90% for each of the four metrics, was found to outperform a rule-base system and a system containing planning alone.

Fostering parent–child dialog through automated discussion suggestions

The development of early literacy skills has been critically linked to a child's later academic success. In particular, repeated studies have shown that reading aloud to children and providing opportunities for them to discuss the stories that they hear is of utmost importance to later academic success. CloudPrimer is a tablet-based interactive reading primer that aims to foster early literacy skills by supporting parents in shared reading with their children through user-targeted discussion topic suggestions. The tablet application records discussions between parents and children as they read a story and, in combination with a common sense knowledge base, leverages this information to produce suggestions. Because of the unique challenges presented by our application, the suggestion generation method relies on a novel topic modeling method that is based on semantic graph topology. We conducted a user study in which we compared how delivering suggestions generated by our approach compares to expert-crafted suggestions. Our results show that our system can successfully improve engagement and parent---child reading practices in the absence of a literacy expert's tutoring.

Modeling content and structure for abstractive review summarization

Reviews are valuable sources of information for many important decision making tasks. Summarizing the massive amount of reviews, which are available these days on many entities and services, is critical to help users better digest the sentiment about an entity or a service and its aspects (i.e. features of the entity or the service). This article presents a novel aspect-based summarization framework that generates an abstract from multiple reviews of an entity without the need for a handcrafted feature taxonomy or any training data. We generate summaries using Natural Language Generation (NLG) by taking into account the importance of aspects, as well as the association between them. We model these information in the form of a tree, called Aspect Hierarchy Tree (AHT), in which nodes indicate the important aspects and edges indicate the relationship between them. We propose and investigate three alternative content selection and structuring models for the automatic construction of an AHT in our summarization framework: 1) Rhetorical model, which captures the aspects' importance and relationship by looking at the way people discuss and relate the aspects when expressing opinion in their reviews. 2) Conceptual model, which exploits a common-sense knowledge base (e.g. ConceptNet) to find the conceptual association between aspects. 3) Hybrid model, which exploits both the rhetorical and conceptual information. Our abstractive summarization framework has the potential to implement one of the proposed models dependingon the application or apply all three models and let a user choose the output, depending on his/her desire to use the conceptual, rhetorical or both sources of information. Quantitative and qualitative analysis on the resulting AHTs of the three content selection and structuring models for seven entities in three domains shows that the three models generate AHTs that differ in interesting ways in terms of both content (i.e. selected aspects to be included in the summary) and structure (i.e. the relation between aspects).

Using a distributional semantic vector space with a knowledge base for reasoning in uncertain conditions

Abstract The inherent inflexibility and incompleteness of commonsense knowledge bases (KB) has limited their usefulness. We describe a system called Displacer for performing KB queries extended with the analogical capabilities of the word2vec distributional semantic vector space (DSVS). This allows the system to answer queries with information which was not contained in the original KB in any form. By performing analogous queries on semantically related terms and mapping their answers back into the context of the original query using displacement vectors, we are able to give approximate answers to many questions which, if posed to the KB alone, would return no results. We also show how the hand-curated knowledge in a KB can be used to increase the accuracy of a DSVS in solving analogy problems. In these ways, a KB and a DSVS can make up for each other’s weaknesses.

Commonsense in Parts: Mining Part-Whole Relations from the Web and Image Tags

Commonsense knowledge about part-whole relations (e.g., screen partOf notebook) is important for interpreting user input in web search and question answering, or for object detection in images. Prior work on knowledge base construction has compiled part-whole assertions, but with substantial limitations: i) semantically different kinds of part-whole relations are conflated into a single generic relation, ii) the arguments of a part-whole assertion are merely words with ambiguous meaning, iii) the assertions lack additional attributes like visibility (e.g., a nose is visible but a kidney is not) and cardinality information (e.g., a bird has two legs while a spider eight), iv) limited coverage of only tens of thousands of assertions. This paper presents a new method for automatically acquiring part-whole commonsense from Web contents and image tags at an unprecedented scale, yielding many millions of assertions, while specifically addressing the four shortcomings of prior work. Our method combines pattern-based information extraction methods with logical reasoning. We carefully distinguish different relations: physicalPartOf, memberOf, substanceOf. We consistently map the arguments of all assertions onto WordNet senses, eliminating the ambiguity of word-level assertions. We identify whether the parts can be visually perceived, and infer cardinalities for the assertions. The resulting commonsense knowledge base has very high quality and high coverage, with an accuracy of 89% determined by extensive sampling, and is publicly available.

영어 작문 자동채점에서 ConceptNet과 작문 프롬프트를 이용한 주제-이탈 문서의 자동 검출

This work presents a new method that can predict, without the use of training data, whether an input essay is written on a given topic. ConceptNet is a common-sense knowledge base that is generated automatically from sentences that are extracted from a variety of document types. An essay prompt is the topic that an essay should be written about. The method that is proposed in this paper uses ConceptNet and an essay prompt to decide whether or not an input essay is off-topic. We introduce a way to find the shortest path between two nodes on ConceptNet, as well as a way to calculate the semantic similarity between two nodes. Not only an essay prompt but also a student's essay can be represented by concept nodes in ConceptNet. The semantic similarity between the concepts that represent an essay prompt and the other concepts that represent a student's essay can be used for a calculation to rank on-topicness ; if a low ranking is derived, an essay is regarded as off-topic. We used eight different essay prompts and a student-essay collection for the performance evaluation, whereby our proposed method shows a performance that is better than those of the previous studies. As ConceptNet enables the conduction of a simple text inference, our new method looks very promising with respect to the design of an essay prompt for which a simple inference is required.

Using association reasoning tool to achieve semantic reframing of service design insight discovery

Developing a desirable service requires in-depth understanding of customers, but discovering customer insights has depended on designers' experiences. It is hard for novice designers and designers at enterprises with Goods Dominant Logic mindsets to do this well. A partial solution rests on challenging the core design practices of reframing and frame creation. This study presents an IT-based tool with the knowledge in ConceptNet (a machine-readable commonsense knowledgebase) to facilitate a variety of association reasoning methods in the reframing and frame creation process. The concept of insight depth is proposed for measuring the influential extent of insights. Our tool has also evidenced its utility on improving the quality of the reframing and frame creation process for experienced and novice designers.

Approximate and selective reasoning on knowledge graphs: A distributional semantics approach

Tasks such as question answering and semantic search are dependent on the ability of querying and reasoning over large-scale commonsense knowledge bases (KBs). However, dealing with commonsense data demands coping with problems such as the increase in schema complexity, semantic inconsistency, incompleteness and scalability. This paper proposes a selective graph navigation mechanism based on a distributional relational semantic model which can be applied to querying and reasoning over heterogeneous knowledge bases (KBs). The approach can be used for approximative reasoning, querying and associational knowledge discovery. In this paper we focus on commonsense reasoning as the main motivational scenario for the approach. The approach focuses on addressing the following problems: (i) providing a semantic selection mechanism for facts which are relevant and meaningful in a specific reasoning and querying context and (ii) allowing coping with information incompleteness in large KBs. The approach is evaluated using ConceptNet as a commonsense KB, and achieved high selectivity, high selectivity scalability and high accuracy in the selection of meaningful navigational paths. Distributional semantics is also used as a principled mechanism to cope with information incompleteness.

혼합 공간 추론 알고리즘의 설계 및 구현

In order to answer questions successfully on behalf of the human contestant in DeepQA environments such as 'Jeopardy!', the American quiz show, the computer needs to have the capability of fast temporal and spatial reasoning on a large-scale commonsense knowledge base. In this paper, we present a hybrid spatial reasoning algorithm, among various efficient spatial reasoning methods, for handling directional and topological relations. Our algorithm not only improves the query processing time while reducing unnecessary reasoning calculation, but also effectively deals with the change of spatial knowledge base, as it takes a hybrid method that combines forward and backward reasoning. Through experiments performed on the sample spatial knowledge base with the hybrid spatial reasoner of our algorithm, we demonstrated the high performance of our hybrid spatial reasoning algorithm.

맵리듀스 프레임워크를 이용한 대용량 공간 추론기의 설계 및 구현

ABSTRACT In order to answer the questions successfully on behalf of the human in DeepQA environments such as Jeopardy! of the American quiz show, the computer is required to have the capability of fast temporal and spatial reasoning on a large-scale commonsense knowledge base. In this paper, we present a scalable spatial reasoning algorithm for deriving efficiently new directional and topological relations using the MapReduce framework, one of well-known parallel distributed computing environments. The proposed reasoning algorithm assumes as input a large-scale spatial knowledge base including CSD-9 directional relations and RCC-8 topological relations. To infer new directional and topological relations from the given spatial knowledge base, it performs the cross-consistency checks as well as the path-consistency checks on the knowledge base. To maximize the parallelism of reasoning computations according to the principle of the MapReduce framework, we design the algorithm to partition effectively the large knowledge base into smaller ones and distribute them over multiple computing nodes at the map phase. And then, at the reduce phase, the algorithm infers the new knowledge from distributed spatial knowledge bases. Through experiments performed on the sample knowledge base with the MapReduce-based implementation of our algorithm, we proved the high performance of our large-scale spatial reasoner.Keywords:Spatial Reasoner, Directional Relations, Topological Relations, Distributed Processing, MapReduce

Using relation selection to improve value propagation in a ConceptNet-based sentiment dictionary

Sentiment analysis has become an important topic in natural language processing in recent years, and sentiment dictionaries are essential for research in this field. Concept-level sentiment dictionaries have broader coverage than word-based dictionaries, but they are still insufficient for real-world applications. In our previous work, we used a commonsense knowledge base (ConceptNet) as the foundation to build a larger dictionary. By propagating sentiment values from concepts with known values to empty concepts, we greatly enlarged our concept-level sentiment dictionary. In this work, we refine our previous method by adding relation selection and bias correction steps. Based on the assumption that concepts pass their sentiment values to their neighbors in different ways depending on the relations connecting them, we use sequential forward search to find the best combination of relations. We also propose a bias correction method that guarantees that the average deviation and standard deviation of sentiment values in the whole sentiment dictionary remain unchanged. We show that the strategy can improve the polarity accuracy by 3.7% and the Kendall τ distance by 17.3% relative to our previous method. Also, our experiment shows that the dictionary we constructed leads to better performance in the sentiment polarity classification task.