Analogical Retrieval Research Articles

Analogical reasoning during hypothesis generation: the effects of object and domain similarities on access and transfer

In two experiments on analogical hypothesis generation, we factorially manipulated the presence of domain and object similarities between a base situation and a target phenomenon, and assessed their effects on the transfer of the source’s explanatory structure before and after an indication to use the base analog as a source for analogical explanations. The absence of any kind of surface similarity led to very low rates of spontaneous transfer. In both experiments, however, either kind of surface similarity sufficed to enhance the spontaneous transfer of the base explanation during the formulation of plausible hypotheses for the target. The transfer advantage of object and domain similarity cannot be attributed to the effect of these variables on post-access processes, since experimental conditions did not differ with regard to the ability to transfer the base explanation upon explicit request. The effect of domain similarity on spontaneous analogical explanation constitutes a relevant finding, especially given the lack of attention received by this dimension of similarity in behavioural studies and computer simulations of analogical retrieval.

Leaping across the mental canyon: higher-order long-distance analogical retrieval

ABSTRACT Previous experiments have shown that a comparison of two written narratives highlights their shared relational structure, which in turn facilitates the retrieval of analogous narratives from the past. However, analogical retrieval occurs across domains that appear more conceptually distant than merely different narratives, and the deepest analogies use matches in higher-order relational structure. The present study investigated whether schema abstraction can facilitate analogical retrieval of higher-order relations across written narratives and abstract symbolic problems. Participants read stories which became retrieval targets after a delay, cued by either analogous stories or letter-strings. We replicated prior research that used narrative retrieval cues, and also found clear evidence that a comparison of analogous letter-string problems and explanation of the underlying principle facilitated the retrieval of source stories with analogous higher-order relations. These findings show that the schemas abstracted from comparison of narratives can be transferred to non-semantic symbolic domains.

SMFM-based analogy retrieval tool for the conceptual design of innovative products

Design-by-Analogy (DbA) is a powerful approach to innovative conceptual design. Cross-domain analogies are the stimulated sources of inspiration for generating new concepts. Therefore, how to efficiently retrieve them becomes an essential issue to be solved. This paper proposes a tool based on a structure-mapping function model (SMFM) for analogy retrieval. Inspired by the structure-mapping theory, SMFM considers two in-depth features: functional relation and causal relation between functions, which are expressed by function and meta-function concepts, respectively. SMFM serves to capture design knowledge from instances (i.e., engineering or biological systems) to establish a case database. The SMFM ontology is constructed for linking knowledge representation and analogy retrieval. Its meta-function, function, and flow terms are used to index the modeled instances so as to establish an index database. Analogy retrieval is realized by the ontology-based query expansion, vector space model and weighted cosine similarity. By inputting function or meta-function queries, we retrieve the required cross-domain analogies, whose design knowledge is displayed by function models to inspire target functions or meta-functions realization. Finally, the tool is evaluated for its effectiveness in cross-domain analogies retrieval. An application example illustrates its practicality in aiding innovative design. Additionally, an experiment was conducted to test the tool. The results indicate that our tool can assist users (e.g., engineering designers) to more quickly generate new schemes to realize target meta-functions and the quality of schemes is higher.

Probabilistic analogical mapping with semantic relation networks.

The human ability to flexibly reason using analogies with domain-general content depends on mechanisms for identifying relations between concepts, and for mapping concepts and their relations across analogs. Building on a recent model of how semantic relations can be learned from nonrelational word embeddings, we present a new computational model of mapping between two analogs. The model adopts a Bayesian framework for probabilistic graph matching, operating on semantic relation networks constructed from distributed representations of individual concepts and of relations between concepts. Through comparisons of model predictions with human performance in a novel mapping task requiring integration of multiple relations, as well as in several classic studies, we demonstrate that the model accounts for a broad range of phenomena involving analogical mapping by both adults and children. We also show the potential for extending the model to deal with analog retrieval. Our approach demonstrates that human-like analogical mapping can emerge from comparison mechanisms applied to rich semantic representations of individual concepts and relations. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

The effect of surface similarities on the retrieval of analogous daily-life events.

Against the typical results of laboratory studies, it has been suggested that retrieving distant analogs from autobiographical memory would be relatively easy, since we frequently encode daily-life events in terms of overlearned relational categories that allow for a uniform abstract encoding. In each of two experiments, we formed two groups of participants who, as determined by a questionnaire presented during a first session, had experienced an event corresponding to a schema-governed category (Experiment 1) or to a system of schema-governed categories (Experiment 2). While the episodes reported by one of the groups belonged to the same domain as the target analog to be presented during the second session, those of the other group belonged to a different domain. During a temporally and contextually separated session, the experimenters presented both groups with a target analog belonging to the schema-governed category for which participants had reported a base analog. Participants had to retrieve an autobiographical episode that they considered analogous to the situation presented by the experimenter. From those analogs reported in the first phase, those pertaining to the same domain of the target were more retrieved than those pertaining to a different domain. Results showed that analogical retrieval is driven largely by surface similarities, even when base and target analogs have been encoded in terms of the same schema-governed category.

Analogy Generation in Science Experts and Novices.

There is a critical inconsistency in the literature on analogical retrieval. On the one hand, a vast set of laboratory studies has found that people often fail to retrieve past experiences that share deep relational commonalities, even when they would be useful for reasoning about a current problem. On the other hand, historical studies and naturalistic research show clear evidence of remindings based on deep relational commonalities. Here, we examine a possible explanation for this inconsistency-namely, that remindings based on relational principles increase as a function of expertise. To test this claim, we devised a simple analogy-generation task that can be administered across a wide range of expertise. We presented common events as the bases from which to generate analogies. Although the events themselves were unrelated to geoscience, we found that when the event was explainable in terms of a causal principle that is prominent in geoscience, expert geoscientists were likely to spontaneously produce analogies from geoscience that relied on the same principle. Further, for these examples, prompts to produce causal analogies increased their frequency among nonscientists and scientists from another domain, but not among expert geoscientists (whose spontaneous causal retrieval levels were already high). In contrast, when the example was best explained by a principle outside of geoscience, all groups required prompting to produce substantial numbers of analogies based on causal principles. Overall, this pattern suggests that the spontaneous use of causal principles is characteristic of experts. We suggest that expert scientists adopt habitual patterns of encoding according to the key relational principles in their domain, and that this contributes to their propensity to spontaneously retrieve relational matches. We discuss implications for the nature of expertise and for science instruction and assessment.

Preventing inert knowledge: Category status promotes spontaneous structure-based retrieval of prior knowledge.

Analogical comparison of 2 provided cases promotes spontaneous analogical transfer by encouraging a more abstract representation of a target principle. This is widely understood as a process of schema abstraction that aids retrieval from memory in the absence of superficial similarity. The category status hypothesis states that if knowledge about a target principle is represented as a relational category, it is easier to activate as a result of categorizing (as opposed to cue-based reminding). To investigate these 2 pathways to spontaneous transfer (schema retrieval vs. categorization), participants were assigned to different study conditions: (a) standard comparison of 2 analogs; (b) standard comparison followed by a second comparison of 2 new analogs; or (c) a guided Category-building task based on a foundation of sequential summarization with additional categorization supports. Experiment 1 showed improved spontaneous transfer in the Category-building task relative to Single-comparison-the gold standard for encouraging spontaneous transfer. Experiment 2 demonstrated that the Category-building task led to more frequent spontaneous structure-based retrieval than either comparison-based task. This suggests that the transfer improvements observed in Experiment 1 are attributable to the Category-building task making superficially dissimilar prior knowledge more accessible. Experiments 3a and 3b tested alternative explanations of this finding and provided evidence that the effectiveness of the Category-building task arises from encouraging the construal of a target principle as a relational category. These findings support categorization as a critical factor in explaining successful spontaneous analogical retrieval and toward instructional approaches that promote portable rather than inert knowledge. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Representation of Characters’ Directed relationships in Comics with Speech-roles

This paper describes a novel way to represent characters’ relationships in comics: as edges of directed graphs labeled with speech-role frequency distribution. In comics, the story progresses through characters’ actions and speeches. The relationships of characters are usually represented as non-directed: e.g., they are “lovers,” or “friends.” However, directionality is an essential component of complicated relationships; the combination of directed relationships often determines the attractiveness of a story. In this study, relationships between characters are assumed to be directed, and are further specified using speech roles: the types of speech that pass between two characters can typify their relationship. This idea has been realized experimentally, and its accuracy has been verified against subjective evaluation by readers. We also suggest the use of directed relationships for analogy retrieval of characters. These methods should be applicable in fields other than comics.

An exploration-based approach to computationally supported design-by-analogy using D3

AbstractComputational support for design-by-analogy (DbA) is a growing field, as it aids the process for designers looking to draw inspiration from external sources by harnessing the power of data mining and data visualization. This study presents a unique exploration-based approach for the analogical retrieval process using a computational tool called VISION (VisualInteraction tool forSeekingInspiration basedOnNonnegative Matrix Factorization). Leveraging the U.S. patent database as a source of inspiration, VISION enables designers to visualize a patent repository and explore for analogical inspiration in a user-driven manner. To achieve this, we perform hierarchical Nonnegative Matrix Factorization to generate a clustered structure of patent data and employ D3.js to visualize the patent structure in a node-link network, in which user interaction capabilities are enabled for data exploration. In this study, we also analyze the effect of data size (ranging from 100 to 3000 patents) on two performance aspects of VISION – the clustering quality of topic modeling results and the frame rate of interactive data visualization. The findings show that the tool exhibits more randomized and inconsistent topic modeling results when the database size is too small. But, increasing the database size lowers the frame rate to the point that it could diminish designers’ ability to retrieve and recall information. The scope of the work here is to present the creation of the DbA visualization tool called VISION and to evaluate its data scale limitations in order to provide a basis for developing a visual interaction tool for the analogical retrieval process during DbA.

Are Superficially Dissimilar Analogs better retrieved than Superficially Similar Disanalogs?

In the present study, we tested the assumption that structural similarity overcomes surface similarity in the retrieval of past events, by observing whether structural similarity alone is a better cue than surface similarity alone. To do so, in four story-recall experiments, we provided the participants with multiple source stories and then with a target cue story. This target cue only shared either surface or structural similarity with the source stories. In Experiment 1A, a Superficially Similar Disanalog source story (SSD) and a Superficially Dissimilar Analog source story (SDA) were presented among Superficially Dissimilar Disanalog source stories (SDDs). A soundness rating task was used in Experiment 1B to control the absence of structural similarity among the SSDs presented in Experiment 1A. In Experiment 2, the number of SSDs was increased in the aim to reproduce more ecological conditions. In two further experiments, a five minute (Experiment 3) and a 45 minute (Experiment 4) delay was introduced, and supplementary source stories were presented, in order to make the study more similar to previous story-recall paradigms. The results of the four story-recall experiments support the dominance of structural over surface similarities in analogical retrieval. The role of a structurally-based access regarding the retrieval of Superficially Similar Analogs (SSAs) and SDAs is discussed, as well as the factors underlying the rare occurrence of SDAs retrievals in previous experiments.

MASS-UMAP: Fast and Accurate Analog Ensemble Search in Weather Radar Archives

The use of analog-similar weather patterns for weather forecasting and analysis is an established method in meteorology. The most challenging aspect of using this approach in the context of operational radar applications is to be able to perform a fast and accurate search for similar spatiotemporal precipitation patterns in a large archive of historical records. In this context, sequential pairwise search is too slow and computationally expensive. Here, we propose an architecture to significantly speed up spatiotemporal analog retrieval by combining nonlinear geometric dimensionality reduction (UMAP) with the fastest known Euclidean search algorithm for time series (MASS) to find radar analogs in constant time, independently of the desired temporal length to match and the number of extracted analogs. We show that UMAP, combined with a grid search protocol over relevant hyperparameters, can find analog sequences with lower mean square error (MSE) than principal component analysis (PCA). Moreover, we show that MASS is 20 times faster than brute force search on the UMAP embedding space. We test the architecture on real dataset and show that it enables precise and fast operational analog ensemble search through more than 2 years of radar archive in less than 3 seconds on a single workstation.

Can a relational mindset boost analogical retrieval?

BackgroundMemory retrieval is driven by similarity between a present situation and some prior experience, but not all similarity is created equal. Analogical retrieval, rooted in the similarity between two situations in their underlying structural relations, is often responsible for new insights and innovative solutions to problems. However, superficial similarity is instead more likely to drive spontaneous retrieval. How can we make analogical retrieval more likely? Inducing a relational mindset via an analogical reasoning task has previously been shown to boost subsequent relational thinking. In this paper, we examined whether inducing a relational mindset could also boost analogical retrieval.ResultsWe find that a relational mindset can increase analogical retrieval if induced before information is encoded in the first place, amplifying the effect of a clearly labelled relational structure. On the other hand, inducing a relational mindset at the time of retrieval did not increase analogical retrieval.ConclusionThis work further demonstrates the central importance of high-quality relational encoding for subsequent relation-based analogical retrieval, and that inducing a relational mindset can improve those encodings.

Design-by-Analogy: Exploring for Analogical Inspiration With Behavior, Material, and Component-Based Structural Representation of Patent Databases

Design-by-analogy (DbA) is an important method for innovation that has gained much attention due to its history of leading to successful and novel design solutions. The method uses a repository of existing design solutions where designers can recognize and retrieve analogical inspirations. Yet, exploring for analogical inspiration has been a laborious task for designers. This work presents a computational methodology that is driven by a topic modeling technique called non-negative matrix factorization (NMF). NMF is widely used in the text mining field for its ability to discover topics within documents based on their semantic content. In the proposed methodology, NMF is performed iteratively to build hierarchical repositories of design solutions, with which designers can explore clusters of analogical stimuli. This methodology has been applied to a repository of mechanical design-related patents, processed to contain only component-, behavior-, or material-based content to test if unique and valuable attribute-based analogical inspiration can be discovered from the different representations of patent data. The hierarchical repositories have been visualized, and a case study has been conducted to test the effectiveness of the analogical retrieval process of the proposed methodology. Overall, this paper demonstrates that the exploration-based computational methodology may provide designers an enhanced control over design repositories to retrieve analogical inspiration for DbA practice.

Evaluating semantic relations in neural word embeddings with biomedical and general domain knowledge bases

BackgroundIn the past few years, neural word embeddings have been widely used in text mining. However, the vector representations of word embeddings mostly act as a black box in downstream applications using them, thereby limiting their interpretability. Even though word embeddings are able to capture semantic regularities in free text documents, it is not clear how different kinds of semantic relations are represented by word embeddings and how semantically-related terms can be retrieved from word embeddings.MethodsTo improve the transparency of word embeddings and the interpretability of the applications using them, in this study, we propose a novel approach for evaluating the semantic relations in word embeddings using external knowledge bases: Wikipedia, WordNet and Unified Medical Language System (UMLS). We trained multiple word embeddings using health-related articles in Wikipedia and then evaluated their performance in the analogy and semantic relation term retrieval tasks. We also assessed if the evaluation results depend on the domain of the textual corpora by comparing the embeddings of health-related Wikipedia articles with those of general Wikipedia articles.ResultsRegarding the retrieval of semantic relations, we were able to retrieve semanti. Meanwhile, the two popular word embedding approaches, Word2vec and GloVe, obtained comparable results on both the analogy retrieval task and the semantic relation retrieval task, while dependency-based word embeddings had much worse performance in both tasks. We also found that the word embeddings trained with health-related Wikipedia articles obtained better performance in the health-related relation retrieval tasks than those trained with general Wikipedia articles.ConclusionIt is evident from this study that word embeddings can group terms with diverse semantic relations together. The domain of the training corpus does have impact on the semantic relations represented by word embeddings. We thus recommend using domain-specific corpus to train word embeddings for domain-specific text mining tasks.

Design by Analogy: Achieving More Patentable Ideas from One Creative Design

A patent is a kind of technical document to protect intellectual property for individuals or enterprises. Patentable idea generation is a crucial step for patent application and analogy is confirmed to be an effective technique to inspire creative ideas. Analogy-based design usually starts from representation of an analogy source and is followed by the retrieval of appropriate analogs, mapping of design knowledge and adaptation of target solution. To diffuse one core idea into other new contexts and achieve more patentable ideas, this paper mainly centered on the first two stages of analogy-based design and proposed a patentable ideation framework. The analogical information of the source system, including source design problems and solution, was mined comprehensively through International Patent Classification analysis and represented in the form of function, behavior and structure. Three heuristics were suggested for searching the set of candidate target systems with a similar design problem, where the source design could be transferred. To systematize the process of source representation, analogs retrieval, idea transfer, and solution generation, an ideation model was put forward. Finally, the bladeless fan was selected as a source design to illustrate the application of this work. The design output shows that the representation and heuristics are beneficial, and this systematic ideation method can help the engineer or designer enhance creativity and discover more patentable opportunities.

Action Recognition From Skeleton Data via Analogical Generalization Over Qualitative Representations

Human action recognition remains a difficult problem for AI. Traditional machine learning techniques can have high recognition accuracy, but they are typically black boxes whose internal models are not inspectable and whose results are not explainable. This paper describes a new pipeline for recognizing human actions from skeleton data via analogical generalization. Specifically, starting with Kinect data, we segment each human action by temporal regions where the motion is qualitatively uniform, creating a sketch graph that provides a form of qualitative representation of the behavior that is easy to visualize. Models are learned from sketch graphs via analogical generalization, which are then used for classification via analogical retrieval. The retrieval process also produces links between the new example and components of the model that provide explanations. To improve recognition accuracy, we implement dynamic feature selection to pick reasonable relational features. We show the explanation advantage of our approach by example, and results on three public datasets illustrate its utility.

Learning From Unannotated QA Pairs to Analogically Disambiguate and Answer Questions

Creating systems that can learn to answer natural language questions has been a longstanding challenge for artificial intelligence. Most prior approaches focused on producing a specialized language system for a particular domain and dataset, and they required training on a large corpus manually annotated with logical forms. This paper introduces an analogy-based approach that instead adapts an existing general purpose semantic parser to answer questions in a novel domain by jointly learning disambiguation heuristics and query construction templates from purely textual question-answer pairs. Our technique uses possible semantic interpretations of the natural language questions and answers to constrain a query-generation procedure, producing cases during training that are subsequently reused via analogical retrieval and composed to answer test questions. Bootstrapping an existing semantic parser in this way significantly reduces the number of training examples needed to accurately answer questions. We demonstrate the efficacy of our technique using the Geoquery corpus, on which it approaches state of the art performance using 10-fold cross validation, shows little decrease in performance with 2-folds, and achieves above 50% accuracy with as few as 10 examples.

Characterizing the Effects of Multiple Analogs and Extraneous Information for Novice Designers in Design-by-Analogy

This study examines how the quantity of ideas and analog transfer in design-by-analogy (DbA) are affected by multiple analogs and extraneous information, or noise, using a between-subjects, factorial experiment. To evaluate the effects of multiple analogs and noise on ideation, the study uses two metrics in conjunction with one another; namely, number of ideas (most typical in engineering design) and recognition of high-level principle (more common in psychology). The quantity analysis included three components: number of ideas generated, number of ideas that use example products (analogs and noise stimuli), and number of ideas that use analogs. The results indicate two important findings: (1) providing multiple analogs during ideation had a positive impact on ideation quantity and analog transfer. Specifically, the number of analog-based ideas increased with increasing number of analogs but eventually reached a “saturation point”; (2) introducing extraneous information (noise) diminished the successful mapping of analogs to design solutions. The presence of extraneous information did not significantly affect student designers' ability to identify high-level principles in analogs. The study demonstrated that some extraneous information was perceived as surface similar analogs. Any design analog retrieval method or automated tool will produce extraneous information, and more work is needed to understand and minimize its impact.

Analogical Retrieval of Folktales: A Cross-Cultural Approach

Abstract This cross-cultural study addressed how individuals retrieve and transfer naturally learned information (i.e., folktales) from long-term memory by analogy with a previously unencountered story, concept, or problem. American and Iranian participants read target stories constructed to be analogous to folktales either familiar or unfamiliar to their culture, all having high structural familiarity and either high or low surface similarity to the source folktales. Participants reported whether targets (analogues) reminded them of any specific folktale they had learned in the past; positive responses plus additional justification (i.e., the folktale’s name or its gist) were interpreted as successful analogical retrievals. The current experiment demonstrated a high overall rate of analogical retrieval for familiar folktales and essentially no retrieval for unfamiliar folktales. There was also reliably more retrieval for analogue stories having higher versus lower surface similarity to target folktales. The high salience of surface similarity was also revealed when participants rated retrieved folktales for similarity to the target. Personal familiarity with folktales increased the retrieval rate, but presenting the folktale’s name as a cue produced mixed effects on retrieval. In summary, individuals readily retrieved culturally familiar folktales from long-term memory when they encountered structurally similar analogues, but retrieval was modulated by surface similarity.

Transforming functional models to critical chain models via expert knowledge and automatic parsing rules for design analogy identification

AbstractCritical chains composed of critical flows and functions have been demonstrated as an effective qualitative analogy retrieval approach based on performance metrics. In prior work, engineers used expert knowledge to transform functional models into critical chain models, which are abstractions of the functional model. Automating this transformation process is highly desirable so as to provide for a robust transformation method. Within this paper, two paradigms for functional modeling abstraction are compared. A series of pruning rules provide an automated transformation approach, and this is compared to the results generated previously through an expert knowledge approach. These two approaches are evaluated against a set of published functional models. The similarity of the resulting transformation of the functional models into critical chain models is evaluated using a functional chain similarity metric, developed in previous work. Once critical chain models are identified, additional model evaluation criteria are used to evaluate the utility of the critical chain models for design analogy identification. Since the functional vocabulary acts as a common language among designers and engineers to abstract and represent critical design artifact information, analogous matching can be made about the functional vocabulary. Thus, the transformation of functional models into critical chain models enables engineers to use functional abstraction as a mechanism to identify design analogies. The critical flow rule is the most effective first step when automatically transforming a functional model to a critical chain model. Further research into more complex critical chain model architectures and the interactions between criteria is merited.