Relational Data Mining Research Articles

Toward Effective Semi-supervised Node Classification with Hybrid Curriculum Pseudo-labeling

Semi-supervised node classification is a crucial challenge in relational data mining and has attracted increasing interest in research on graph neural networks (GNNs). However, previous approaches merely utilize labeled nodes to supervise the overall optimization, but fail to sufficiently explore the information of their underlying label distribution. Even worse, they often overlook the robustness of models, which may cause instability of network outputs to random perturbations. To address the aforementioned shortcomings, we develop a novel framework termed Hybrid Curriculum Pseudo-Labeling (HCPL) for efficient semi-supervised node classification. Technically, HCPL iteratively annotates unlabeled nodes by training a GNN model on the labeled samples and any previously pseudo-labeled samples, and repeatedly conducts this process. To improve the model robustness, we introduce a hybrid pseudo-labeling strategy that incorporates both prediction confidence and uncertainty under random perturbations, therefore mitigating the influence of erroneous pseudo-labels. Finally, we leverage the idea of curriculum learning to start from annotating easy samples, and gradually explore hard samples as the iteration grows. Extensive experiments on a number of benchmarks demonstrate that our HCPL beats various state-of-the-art baselines in diverse settings.

Oncogenic Gαq activates RhoJ through PDZ-RhoGEF

Oncogenic Gαq causes uveal melanoma via non-canonical signaling pathways. This constitutively active mutant GTPase is also found in cutaneous melanoma, lung adenocarcinoma, and seminoma, as well as in benign vascular tumors, such as congenital hemangiomas. We recently described that PDZ-RhoGEF (also known as ARHGEF11), a canonical Gα12/13 effector, is enabled by Gαs Q227L to activate CdcIn addition, and we demonstrated that constitutively active Gαq interacts with the PDZ-RhoGEF DH-PH catalytic module, but does not affect its binding to RhoA or Cdc. This suggests that it guides this RhoGEF to gain affinity for other GTPases. Since RhoJ, a small GTPase of the Cdc42 subfamily, has been involved in tumor-induced angiogenesis and the metastatic dissemination of cancer cells, we hypothesized that it might be a target of oncogenic Gαq signaling via PDZ-RhoGEF. Consistent with this possibility, we found that Gαq Q209L drives full-length PDZ-RhoGEF and a DH-PH construct to interact with nucleotide-free RhoJ-G33A, a mutant with affinity for active RhoJ-GEFs. Gαq Q209L binding to PDZ-RhoGEF was mapped to the PH domain, which, as an isolated construct, attenuated the interaction of this mutant GTPase with PDZ-RhoGEF’s catalytic module (DH-PH domains). Expression of these catalytic domains caused contraction of endothelial cells and generated fine cell sprouts that were inhibited by co-expression of dominant negative RhoJ. Using relational data mining of uveal melanoma patient TCGA datasets, we got an insight into the signaling landscape that accompanies the Gαq/PDZ-RhoGEF/RhoJ axis. We identified three transcriptional signatures statistically linked with shorter patient survival, including GPCRs and signaling effectors that are recognized as vulnerabilities in cancer cell synthetic lethality datasets. In conclusion, we demonstrated that an oncogenic Gαq mutant enables the PDZ-RhoGEF DH-PH module to recognize RhoJ, suggesting an allosteric mechanism by which this constitutively active GTPase stimulates RhoJ via PDZ-RhoGEF. These findings highlight PDZ-RhoGEF and RhoJ as potential targets in tumors driven by mutant Gαq.

Qualitative motivation with sets and relations.

In self-determination theory (SDT), multiple conceptual regulations of motivation are posited. These forms of motivation are especially qualitatively viewed by SDT researchers, and there are situations in which combinations of these regulations occur. In this article, instead of the commonly used numerical approach, this is modeled more versatilely by sets and relations. We discuss discrete mathematical models from the theory of knowledge spaces for the combinatorial conceptualization of motivation. Thereby, we constructively add insight into a dispute of opinions on the unidimensionality vs. multidimensionality of motivation in SDT literature. The motivation order derived in our example, albeit doubly branched, was approximately a chain, and we could quantify the combinatorial details of that approximation. Essentially, two combinatorial dimensions reducible to one were observed, which could be studied in other more popular scales as well. This approach allows us to define the distinct, including even equally informative, gradations of any regulation type. Thus, we may identify specific forms of motivation that may otherwise be difficult to measure or not be separable empirically. This could help to dissolve possible inconsistencies that may arise in applications of the theory in distinguishing the different regulation types. How to obtain the motivation structures in practice is demonstrated by relational data mining. The technique applied is an inductive item tree analysis, an established method of Boolean analysis of questionnaires. For a data set on learning motivation, the motivation spaces and co-occurrence relations for the gradations of the basic regulation types are extracted, thus, enumerating their potential subforms. In that empirical application, the underlying models were computed within each of the intrinsic, identified, introjected, and external regulations, in autonomous and controlled motivations, and the entire motivation domain. In future studies, the approach of this article could be employed to develop adaptive assessment and training procedures in SDT contexts and for dynamical extensions of the theory, if motivational behavior can go in time.

Acupuncture and Moxibustion Treatment of Mild Cognitive Impairment Based on Data Mining

Acupuncture and moxibustion therapy, as a traditional medical treatment method of the motherland, is widely used in the treatment of mild cognitive dysfunction, with remarkable curative effect, and is increasingly recognized by the medical community. However, so far, there is no acupuncture prescription and acupoint usage rules for acupuncture treatment of mild cognitive impairment. Therefore, it is of great significance to collect, organize and data mining related data in order to summarize the relationship between acupuncture points and mild cognitive dysfunction, and to provide references for clinical treatment. This article standardizes the acupuncture points in the database. This paper uses Excel and SPS software to carry out frequency analysis, cluster analysis and association rules, and draws the rules of acupoints, meridians, positions, frequency of specific acupoints, clustering of acupoints and compatibility. In this paper, the number of cases in the treatment group and the control group, the main acupoints and matching points of the acupuncture prescription, the treatment methods, the efficacy evaluation indicators, and the total effective rate of the treatment group and the control group are extracted into the Excel table to form a mild cognition of acupuncture treatment Prescription database for dysfunction. This article studies the rules and characteristics of acupuncture points derived from data mining, takes the general rules and characteristics of clinical acupuncture and moxibustion treatment as a reference for clinical acupuncture and moxibustion, and verifies it with the results obtained. The experimental results show that in the analysis of association rules, taking Baihui and Shenting as examples, the support degree indicates that the probability that the two acupoints appear in the 126 heads at the same time is 51.83%, and the confidence and support for the occurrence of Baihui and Shenting are relatively high. High, it is considered to be a common compatible point pair for acupuncture treatment of mild cognitive dysfunction.

Adaptive Hypergraph Auto-Encoder for Relational Data Clustering

The embedded representation and clustering tasks both play important roles in relational data analysis and mining. Traditional methods mainly employ graph structure to describe relational data, but intuitive pairwise connections among nodes are insufficient to model high-order data in the real-world, such as the relations between proteins and polypeptide chains. Hypergraphs are a generalization of graphs, and hypergraphs can well model high-order data. When modeling relational data in the real world, hypergraphs are often accompanied by node attributes, i.e. attributed hypergraphs. Besides this, how to integrate the structural information and attribute information appropriately is another important task, while has not been investigated systematically. In this paper, we propose Adaptive Hypergraph Auto-Encoder(AHGAE) to learn node embeddings in low-dimensional space. Our method can utilize the high-order relation to generate embedding for clustering. It is composed of two procedures, i.e. the adaptive hypergraph Laplacian smoothing filter and the relational reconstruction auto-encoder. It has the advantage of integrating more complex data relations compared with graph-based methods, which leads to better modeling and clustering performance. The proposed method has been evaluated on hypergraph datasets and benchmark graph datasets. Experimental results and comparison with the state-of-the-art methods have demonstrated the effectiveness of our proposed method.

A comparison of characteristic parameters of mining related and tectonic seismic aftershock sequences

The temporal decay pattern of seismicity following rock bursts and large magnitude events is studied for aftershock sequences in several mines in Ontario, Canada. The decay phenomenon is described by the so – called Modified Omori’s law. The dependency of the parameters K, c and p with the magnitude of large mine seismic events is analyzed. A total of eleven mining related seismic sequences, with mainshock magnitudes ranging from 0.5 to 4.0 Mn (Nuttli magnitude scale) and from four different mine environments are examined. Results, including the calculation of the Gutenberg - Richter’s and Reasenberg – Jones parameters b and a’, are compared with previous research in tectonic seismicity using available data from New Zealand, Italy and U.S.A. Further parameters were derived and applied in a correlation study of mining related and tectonic events. The study revealed a good correlation between the productivity parameter a’ and b – values. Particularly, the b – values show significant different number ranges for mining related and tectonic data, which was applied to derive a direct numerical relationship between mining related and tectonic parameters and the aftershock number estimation. Finally, numerical values for the b – values in mining related seismic activity were obtained.

Improving mine-to-mill by data warehousing and data mining

ABSTRACTMining is an interdisciplinary industry that utilises equipment and technology intensively in daily operations. Mine-to-Mill is considered as a key concept for metal mining recently. Impact of underperformed basic upstream operations such as drilling and blasting will sustain this inefficiency in downstream processes, such as mineral processing. Data provided for each of these operations from software and hardware utilised on field reached a level where advanced data analytics becomes applicable. Data warehousing and data mining are alternative tools that rely on a robust data structure. This study gives insight into a data-driven framework for modern mines and presents a data mining implementation on real-time mining-related data for prediction of blasting performance. Random forest and adaptive boosting algorithm were utilised on an integrated data warehouse to discover major operational parameters for efficient blasting. The implementation on site improved the performance of drilling and blasting. The variables highlighted as important by random forest and adaptive boosting algorithm directed the experts of mine-to-mill on site to focus on the close control and detailed analysis of certain drilling- and blasting-related parameters.

A scalable robust and automatic propositionalization approach for Bayesian classification of large mixed numerical and categorical data

Companies want to extract value from their relational databases. This is the aim of relational data mining. Propositionalization is one possible approach to relational data mining. Propositionalization adds new attributes, called features, to the main table, leading to an attribute-value representation, a single table, on which a propositional learner can be applied. However, current relational databases are large and composed of mixed, numerical and categorical, data. Moreover, the specificity of relational data is to involve one-to-many relationships. As an example of such data, consider customers purchasing products: each customer can purchase several products. Therefore, there is a need for techniques able to learn complex aggregates. Learning such features means to explore a combinatorial, possibly infinite, space and such an approach is prone to overfitting. We introduce a propositionalization approach dedicated to a robust Bayesian classifier. It efficiently samples a given number of features in the language bias, following a distribution over the complex aggregates. This distribution is also used to penalize complex aggregates in the regularization of the robust Bayesian classifier. Experiments show that it performs better than state-of-the-art methods on most investigated benchmarks and can deal with large datasets more easily. A new real, large, mixed relational dataset is introduced which confirms the ability of our approach to learn complex aggregates.

Recent granular computing frameworks for mining relational data

A lot of data currently being collected is stored in databases with a relational structure. The process of knowledge discovery from such data is a more challenging task compared with single table data. Granular computing, which has successfully been applied to mining data storable in single tables, is a promising direction for discovering knowledge from relational data. This paper summarizes some recent developments in the area of application of granular computing to mining relational data. Four granular computing frameworks for processing relational data are introduced and compared. The paper shows how each of the frameworks represents relational data, constructs information granules and build patterns based on the granules. A generic system that can employ any of the frameworks to discover knowledge from relational data is also outlined.

Visually Mining Relational Data

Mining relational data often boils down to computing clusters, that is finding sub-communities of data elements forming cohesive sub-units, while being well separated from one another. The clusters themselves are sometimes termed “communities” and the way clusters relate to one another is often referred to as a “community structure”. Methods for identifying communities or subgroups in network data is the focus of intense research is different scientific communities and for different purposes. The present paper focuses on two novel algorithms producing multilevel community structures from raw network data. The two algorithms exploit an edge metric extending Watts's clustering coefficient to edges of a graph. The full benefit of the method comes from the multilevel nature of the community structure as it facilitates the visual interaction and navigation of the network by zooming in and out of components at any level. This multilevel navigation proves to be useful when visually exploring a network in search for structural patterns.

Learning Analytics, the Academic Library, and Positive Intellectual Freedom

College and universities are actively developing capacity for learning analytics, which is a sociotechnical system supported by an assemblage of educational data mining technologies and related Big Data practices. Like many Big Data technologies, learning analytics implicates privacy by surveilling behaviors captured by data-based systems and aggregating and analyzing personal information. Issues of privacy are often linked to concerns about intellectual freedom. Consequently, librarians fervently argue for surveillance-free spaces and places to promote the conditions they believe are necessary to support intellectual freedom. In this commentary, I contrast this common view of intellectual freedom with a separate theory, “positive intellectual freedom,” to show how libraries may be able to participate in learning analytics practices while upholding intellectual freedom as a lodestar guiding practice and policy.

Properties of a Granular Computing Framework for Mining Relational Data

This work investigates properties of a framework for mining relational data. The framework is constructed based on granular computing theory and is equipped with a method for deriving information granules from relational data. Such granules are the basis for discovering knowledge of a different type. It is shown in the paper that thanks to the properties one can improve the performance of tasks such as relational objects representation, search space limitation, and relational patterns generation.

Visualising changing tenure relationships: the talking titler methodology, data mining and social network analysis

Networks of social relationships and how they evolve in relation to land interests are a fundamental attribute of land tenure systems. Individuals may be members of complex, multi-layered, changing and conflicting social networks at any particular time in changing situations. The Talking Titler relational data model (TTM), data mining and social network analysis (SNA) were used to visualise and explore rapidly changing land interests using an urbanizing customary society as the illustrative context. Land tenure relationships may change rapidly in situations where both external forces and internal agents foment conflict and social change in local politics, and elites may abuse their power to grab land. Peri-urban customary systems evolve in response to urbanization, titling programmes and land sales that put pressure on traditional customary norms. Social network analysis can facilitate visualizing complex patterns of relationships between people and how they relate to land. It may also identify new, previous, unknown and changing relationships and structures on the ground. There are practical challenges to applying the tools, but the methodology and data mining and network analysis techniques show promise as an addition to the set of tools that may be used to improve tenure security and as a scenario analysis tool.

Probabilistic relational model benchmark generation: Principle and application

The validation of any database mining methodology goes through an evaluation process where benchmarks availability is essential. In this paper, we aim to randomly generate relational database benchmarks that allow to check probabilistic dependencies among the attributes. We are particularly interested in Probabilistic relational models (PRMs). These latter extend Bayesian networks (BNs) to a relational data mining context that enable effective and robust reasoning about relational data structures. Even though a panoply of works have focused, separately, on Bayesian networks and relational databases random generation, no work has been identified for PRMs on that track. This paper provides an algorithmic approach allowing to generate random PRMs from scratch to cover the absence of generation process. The proposed method allows to generate PRMs as well as synthetic relational data from a randomly generated relational schema and a random set of probabilistic dependencies. This can be of interest for machine learning researchers to evaluate their proposals in a common framework, as for databases designers to evaluate the effectiveness of the components of a database management system.

Compound approximation spaces for relational data

• Compound approximation spaces for relational data are proposed. • They are viewed as expansions of tolerance approximation spaces to a relational case. • Properties of compound approximation spaces are investigated. • Compound approximation spaces enable to derive new knowledge from relational data. Rough set theory provides a powerful tool for dealing with uncertainty in data. Application of variety of rough set models to mining data stored in a single table has been widely studied. However, analysis of data stored in a relational structure using rough sets is still an extensive research area. This paper proposes compound approximation spaces and their constrained versions that are intended for handling uncertainty in relational data. The proposed spaces are expansions of tolerance approximation ones to a relational case. Compared with compound approximation spaces, the constrained version enables to derive new knowledge from relational data. The proposed approach can improve mining relational data that is uncertain, incomplete, or inconsistent.

An efficient multi relational framework using fuzzy rule-based classification technique

Multi relational classification is one of the rapidly rising subclasses of multi relational data mining. It focuses on identifying attractive patterns in an interconnected multiple database relations via primary key/foreign key relationship. Many conventional rule-based classification approaches produce the rules from single 'flat' relation and typically suffer from rule clashes. However, many real world data repositories are stored in multiple database relations. Owing to the increase in the quantity of relational data that are being accumulated and the limitation of propositional rule-based classifiers in relational domains, multi relational classification has turned out to be a field with significance. In this work, a multi relational fuzzy rule system (MRFRS) is proposed for classification across multiple database relations based on PART algorithm that generates crisp rules. These rules are used to formulate the multi relational fuzzy rule system that produces rules of mutually exclusive and exhaustive in nature. The empirical results show that the proposed method outperforms well on real datasets in terms of efficiency and accuracy.

Relation-based granules to represent relational data and patterns

The complex structure of relational data makes the process of knowledge discovery from data a more challenging task compared with the single table data structure. The usefulness of granular computing based approaches to mining data stored in a single table is a driving force for adapting this method to relational data. This paper proposes relation-based granules that are defined in a granular computing based approach to mining relational data. The relations are used to represent relational data and patterns to be discovered. Thanks to this representation, the generation of patterns can be speeded up. The representation also makes it possible to discover richer knowledge from relational data.

Flexible propositionalization of continuous attributes in relational data mining

In a relational database, data are stored in primary and secondary tables. Propositionalization can transform a relational database into a single attribute-value table, and hence becomes a useful technique for mining relational databases. However, most of the existing propositionalization approaches deal with categorical attributes, and cannot handle a threshold on an attribute and a threshold on the number of objects satisfying the condition on the attribute at the same time. In this paper, we propose a new propositionalization technique called Cardinalization to solve these problems. In order to handle relative numbers, we propose a second variant of our approach called Quantiles which can discretize the cardinality of Cardinalization and achieve a fixed number of features. Therefore, the Quantiles method can be tuned to different deployment contexts. Additionally, we often observe that the best combination of propositionalization and classification methods depends on the new context (e.g., online/incremental learning). One effective solution could be to predict the optimal combination at training time and use it in different deployment contexts. Here we also propose an effective wrapping algorithm, called WPC (Wrapper to combine Propositionalizer and Classifier) to select the best combination of propositionalization and classification methods to address this task. Extensive performance analyses in synthetic and real-life datasets show that our approach is very effective and efficient in relational data mining.

Wordification: Propositionalization by unfolding relational data into bags of words

Inductive Logic Programming (ILP) and Relational Data Mining (RDM) address the task of inducing models or patterns from multi-relational data. One of the established approaches to RDM is propositionalization, characterized by transforming a relational database into a single-table representation. This paper presents a propositionalization technique called wordification which can be seen as a transformation of a relational database into a corpus of text documents. Wordification constructs simple, easy to understand features, acting as words in the transformed Bag-Of-Words representation. This paper presents the wordification methodology, together with an experimental comparison of several propositionalization approaches on seven relational datasets. The main advantages of the approach are: simple implementation, accuracy comparable to competitive methods, and greater scalability, as it performs several times faster on all experimental databases. Furthermore, the wordification methodology and the evaluation procedure are implemented as executable workflows in the web-based data mining platform ClowdFlows. The implemented workflows include also several other ILP and RDM algorithms, as well as the utility components that were added to the platform to enable access to these techniques to a wider research audience.

Exploration of a Large Database of French Notarial Acts with Social Network Methods

This article illustrates how mathematical and statistical tools designed to handle relational data may be useful to help decipher the most important features and defects of a large historical database and to gain knowledge about a corpus made of several thousand documents. Such a relational model is generally enough to address a wide variety of problems, including most databases containing relational tables. In mathematics, it is referred to as a network or a graph. The article's purpose is to emphasise how a relevant relational model of a historical corpus can serve as a theoretical framework which makes available automatic data mining methods designed for graphs. By such methods, for one thing, consistency checking can be performed so as to extract possible transcription errors or interpretation errors during the transcription automatically. Moreover, when the database is so large that a human being is unable to gain much knowledge by even an exhaustive manual exploration, relational data mining can help elucidate the database's main features. First, the macroscopic structure of the relations between entities can be emphasised with the help of network summaries automatically produced by classification methods. A complementary point of view is obtained via local summaries of the relation structure: a set of network-related indicators can be calculated for each entity, singling out, for instance, highly connected entities. Finally, visualisation methods dedicated to graphs can be used to give the user an intuitive understanding of the database. Additional information can be superimposed on such network visualisations, making it possible to intuitively link the relations between entities using attributes that describe each entity. This overall approach is here illustrated with a large corpus of medieval notarial acts, containing several thousand transactions and involving a comparable number of persons.