Large Raw Data Research Articles

(Invited) Towards High-Performance Domain-Specific in-Sensor and in-Memory Accelerators

Internet of Things (IoT) devices are projected to attain an $1100B market by 2025, with a web of interconnection projected to comprise approximately 75+ billion IoT devices. The large number of IoTs consist of sensory systems that enable massive data collection from the environment and people. However, considerable portions of the captured sensory data are redundant and unstructured. Data conversion of such large raw data, storing in volatile memories, transmission, and computation in on-/off-chip processors, impose high energy consumption, latency, and a memory bottleneck at the edge. Therefore, high-speed, low-power, and normally-off computing domain-specific architectures should be explored and developed to overcome these issues. Motivated by the aforementioned concerns, we will be focusing on cross-layer (device/circuit/architecture/application) co-design of energy-efficient and high-performance processing-in-sensor and processing-in-memory platforms for implementing complex AI and machine learning tasks, bioinformatics tasks, graph processing, etc. We explain how to leverage innovations from circuits and architecture to integrate sensors, memory, and logic to break the existing memory and power walls.

Executing Ad-Hoc Queries on Large Geospatial Data Sets Without Acceleration Structures

In this case study, we investigate if it is possible to harness the capabilities of modern commodity hardware to perform ad-hoc queries on large raw geospatial data sets. Normally, this requires building an index structure, which is a time-consuming process. We aim to provide means to individual users who receive a new or updated geospatial data set and want to directly start working with it without having to build such an index structure first. To this end, we conduct various experiments on two distinct types of data: 3D building models and point clouds. For the former, we demonstrate that well-known algorithms such as fast string search allow a wide range of queries to be answered in at most a few seconds on data sets with over a million buildings. The usage of progressive indexing additionally improves query run time by more than a factor of two. Regarding point clouds, we achieve similar run times using the popular LAS file format and a query throughput of up to a billion points per second when using a columnar memory layout. The run time of ad-hoc queries is often on par with that of database-driven solutions, sometimes even outperforming them. Considering that ad-hoc queries require no preprocessing, our results show that they are a viable alternative to acceleration structures when working with geospatial data.

Learning Dense Features for Point Cloud Registration Using a Graph Attention Network

Point cloud registration is a fundamental task in many applications such as localization, mapping, tracking, and reconstruction. Successful registration relies on extracting robust and discriminative geometric features. Though existing learning-based methods require high computing capacity for processing a large number of raw points at the same time, computational capacity limitation is not an issue thanks to powerful parallel computing process using GPU. In this paper, we introduce a framework that efficiently and economically extracts dense features using a graph attention network for point cloud matching and registration (DFGAT). The detector of the DFGAT is responsible for finding highly reliable key points in large raw data sets. The descriptor of the DFGAT takes these keypoints combined with their neighbors to extract invariant density features in preparation for the matching. The graph attention network (GAT) uses the attention mechanism that enriches the relationships between point clouds. Finally, we consider this as an optimal transport problem and use the Sinkhorn algorithm to find positive and negative matches. We perform thorough tests on the KITTI dataset and evaluate the effectiveness of this approach. The results show that this method with the efficiently compact keypoint selection and description can achieve the best performance matching metrics and reach the highest success ratio of 99.88% registration in comparison with other state-of-the-art approaches.

Context aware benchmarking and tuning of a TByte-scale air quality database and web service

We present context-aware benchmarking and performance engineering of a mature TByte-scale air quality database system which was created by the Tropospheric Ozone Assessment Report (TOAR) and contains one of the world’s largest collections of near-surface air quality measurements. A special feature of our data service https://join.fz-juelich.de is on-demand processing of several air quality metrics directly from the TOAR database. As a service that is used by more than 350 users of the international air quality research community, our web service must be easily accessible and functionally flexible, while delivering good performance. The current on-demand calculations of air quality metrics outside the database together with the necessary transfer of large volume raw data are identified as the major performance bottleneck. In this study, we therefore explore and benchmark in-database approaches for the statistical processing, which results in performance enhancements of up to 32%.

Big data and Smart data: two interdependent and synergistic digital policies within a virtuous data exploitation loop

This research examines for the first time the relationship between Big data and Smart data among French automotive distributors. Many low-tech firms engage in these data policies to improve their decisions and performance through the predictive capacities of their data. A discussion emerges in the literature according to which an effective policy lies in the conversion of a mass of raw data into so-called intelligent data. In order to understand better this digital transition, we question the transformation of data policies practiced in low-tech firms through the founding model of 3Vs (Volume, Variety and Velocity of data). First of all, this empirical study of 112 French automotive distributors develops the existing literature by proposing an original and detailed typology of the data policies practiced (Low data, Big data and Smart data). Secondly, after specifying the elements of the differences between the quantitative nature of Big data and the qualitative nature of Smart data, our results reveal and analyse for the first time the existence of their synergistic relationship. Companies transform their Big data approach into Smart data when they move from massive exploitation to intelligent exploitation of their data. The phenomenon is part of a high-end loop data exploitation. Initially, the exploitation of intelligent data can only be done by extracting a sample from a large raw data pool previously made by a Big data policy. Secondly, the organization's raw data pool is in turn enriched by the repayment of contributions made by the Smart data approach. Thus, this study develops three important ways. First off, we identify, detail and compare the current data policies of a traditional industry. Secondly, we reveal and explain the evolution of digital practices within organizations that now combine both quantitative and qualitative data exploitation. Finally, our results guide decision-makers towards the synergistic and the legitimate association of different forms of data management for better performance.

Data science for building energy efficiency: A comprehensive text-mining driven review of scientific literature

The ever-changing data science landscape is fueling innovation in the built environment context by providing new and more effective means of converting large raw data sets into value for professionals in the design, construction and operations of buildings. The literature developed due to this convergence has rapidly increased in recent years, making it difficult for traditional review approaches to cover all related papers. Therefore, this paper applies a natural language processing (NLP) method to provide an exhaustive and quantitative review.Approximately 30,000 scientific publications were retrieved from the Elsevier API to extract the relationship between data sources, data science techniques, and building energy efficiency applications across the life cycle of buildings. The text-mining and NLP analysis reveals that data sciences techniques are applied more for operation phase applications such as fault detection and diagnosis (FDD), while being under-explored in design and commissioning phases. In addition, it is pointed out that more data science techniques that are to be investigated for various applications. For example, generative adversarial networks (GANs) has potential in facilitating parametric design; transfer learning is a promising path to promoting the application of optimal building operation;

Mapping the entire functionally active endometrial microbiota.

Does endometrium harbour functionally active microorganisms and whether the microbial composition differs between proliferative and mid-secretory phases? Endometrium harbours functionally alive microorganisms including bacteria, viruses, archaea and fungi whose composition and metabolic functions change along the menstrual cycle. Resident microbes in the endometrium have been detected, where microbial dysfunction has been associated with reproductive health and disease. Nevertheless, the core microorganismal composition in healthy endometrium is not determined and whether the identified bacterial DNA sequences refer to alive/functionally active microbes is not clear. Furthermore, whether there are cyclical changes in the microbial composition remains an open issue. RNA sequencing (RNAseq) data from 14 endometrial paired samples from healthy women, 7 samples from the mid-secretory phase and 7 samples from the consecutive proliferative phase were analysed for the microbial RNA sequences. The raw RNAseq data were converted into FASTQ format using SRA Toolkit. The unmapped reads to human sequences were aligned to the reference database Kraken2 and visualised with Krona software. Menstrual phase taxonomic differences were performed by R package metagenomeSeq. The functional analysis of endometrial microbiota was obtained with HUMANn2 and the comparison between menstrual phases was conducted by one-way ANOVA. Human RNAseq analysis was performed using miARma-Seq and the functional enrichment analysis was carried out using gene set enrichment analysis (GSEA; HumanCyc). The integration of metabolic pathways between host and microbes was investigated. The developed method of active microbiota mapping was validated in independent sample set. With the novel metatranscriptomic approach, we mapped the entire alive microbiota composing of >5300 microorganisms within the endometrium of healthy women. Microbes such as bacteria, fungi, viruses and archaea were identified. The validation of three independent endometrial samples from different ethnicity confirmed the findings. Significant differences in the microbial abundances in the mid-secretory vs. proliferative phases were detected with possible metabolic activity in the host-microbiota crosstalk in receptive phase endometrium, specifically in the prostanoid biosynthesis pathway and L-tryptophan metabolism. The raw RNAseq data used in the current study are available at GEO GSE86491 and at BioProject PRJNA379542. These pioneering results should be confirmed in a bigger sample size. Our study confirms the presence of active microbes, bacteria, fungi, viruses and archaea in the healthy human endometrium with implications in receptive phase endometrial functions, meaning that microbial dysfunction could impair the metabolic pathways important for endometrial receptivity. The results of this study contribute to the better understanding of endometrial microbiota composition in healthy women and its possible role in endometrial functions. In addition, our novel methodological pipeline for analysing alive microbes with transcriptional and metabolic activities could serve to inspire new analysis approaches in reproductive medicine. This work is supported by the Spanish Ministry of Economy, Industry and Competitiveness (MINECO) and European Regional Development Fund (FEDER): grants RYC-2016-21199 and ENDORE SAF2017-87526-R; FEDER/Junta de Andalucía-Consejería de Economía y Conocimiento: MENDO (B-CTS-500-UGR18) and by the University of Granada Plan Propio de Investigación 2016 - Excellence actions: Unit of Excellence on Exercise and Health (UCEES) (SOMM17/6107/UGR). A.S.-L. and N.M.M. are funded by the Spanish Ministry of Science, Innovation and Universities (PRE2018-0854409 and FPU19/01638). S.A. has received honoraria for lectures from Merck. The funder had no role in this study.

BnaGVD: A genomic variation database of rapeseed (Brassica napus).

Rapeseed (Brassica napus L.) is a typical polyploid crop and one of the most important oilseed crops worldwide. With the rapid progress on high-throughput sequencing technologies and the reduction of sequencing cost, large-scale genomic data of a specific crop have become available. However, raw sequence data are mostly deposited in the sequence read archive of the National Center of Biotechnology Information (NCBI) and the European Nucleotide Archive (ENA), which is freely accessible to all researchers. Extensive tools for practical purposes should be developed to efficiently utilize these large raw data. Here, we report a web-based rapeseed genomic variation database (BnaGVD, http://rapeseed.biocloud.net/home) from which genomic variations, such as single nucleotide polymorphisms (SNPs) and insertions/deletions (InDels) across a world-wide collection of rapeseed accessions, can be referred. The current release of the BnaGVD contains 34,591,899 high-quality SNPs and 12,281,923 high-quality InDels and provides search tools to retrieve genomic variations and gene annotations across 1,007 accessions of worldwide rapeseed germplasm. We implement a variety of built-in tools (e.g., BnaGWAS, BnaPCA, and BnaStructure) to help users perform in-depth analyses. We recommend this web resource for accelerating studies on the functional genomics and screening of molecular markers for rapeseed breeding.

FIViz: Forensics Investigation through Visualization for Malware in Internet of Things

Adoption of the Internet of Things for the realization of smart cities in various domains has been pushed by the advancements in Information Communication and Technology. Transportation, power delivery, environmental monitoring, and medical applications are among the front runners when it comes to leveraging the benefits of IoT for improving services through modern decision support systems. Though with the enormous usage of the Internet of Medical Things, security and privacy become intrinsic issues, thus adversaries can exploit these devices or information on these devices for malicious intents. These devices generate and log large and complex raw data which are used by decision support systems to provide better care to patients. Investigation of these enormous and complicated data from a victim’s device is a daunting and time-consuming task for an investigator. Different feature-based frameworks have been proposed to resolve this problem to detect early and effectively the access logs to better assess the event. But the problem with the existing approaches is that it forces the investigator to manually comb through collected data which can contain a huge amount of irrelevant data. These data are provided normally in textual form to the investigators which are too time-consuming for the investigations even if they can utilize machine learning or natural language processing techniques. In this paper, we proposed a visualization-based approach to tackle the problem of investigating large and complex raw data sets from the Internet of Medical Things. Our contribution in this work is twofold. Firstly, we create a data set through a dynamic behavioral analysis of 400 malware samples. Secondly, the resultant and reduced data set were then visualized most feasibly. This is to investigate an incident easily. The experimental results show that an investigator can investigate large amounts of data in an easy and time-efficient manner through the effective use of visualization techniques.

In-situ visual exploration over big raw data

Data exploration and visual analytics systems are of great importance in Open Science scenarios, where less tech-savvy researchers wish to access and visually explore big raw data files (e.g., json, csv) generated by scientific experiments using commodity hardware and without being overwhelmed in the tedious processes of data loading, indexing and query optimization. In this paper, we present our work for enabling efficient query processing on large raw data files for interactive visual exploration scenarios and analytics. We introduce a framework, named RawVis, built on top of a lightweight in-memory tile-based index, VALINOR, that is constructed on-the-fly given the first user query over a raw file and progressively adapted based on the user interaction. We evaluate the performance of a prototype implementation compared to three other alternatives and show that our method outperforms in terms of response time, disk accesses and memory consumption. Particularly during an exploration scenario, the proposed method in most cases is about 5-10× faster compared to existing solutions, and requires significantly less memory resources.

Evaluating Commonsense in Pre-Trained Language Models

Contextualized representations trained over large raw text data have given remarkable improvements for NLP tasks including question answering and reading comprehension. There have been works showing that syntactic, semantic and word sense knowledge are contained in such representations, which explains why they benefit such tasks. However, relatively little work has been done investigating commonsense knowledge contained in contextualized representations, which is crucial for human question answering and reading comprehension. We study the commonsense ability of GPT, BERT, XLNet, and RoBERTa by testing them on seven challenging benchmarks, finding that language modeling and its variants are effective objectives for promoting models' commonsense ability while bi-directional context and larger training set are bonuses. We additionally find that current models do poorly on tasks require more necessary inference steps. Finally, we test the robustness of models by making dual test cases, which are correlated so that the correct prediction of one sample should lead to correct prediction of the other. Interestingly, the models show confusion on these test cases, which suggests that they learn commonsense at the surface rather than the deep level. We release a test set, named CATs publicly, for future research.

PSO Based Grey Bass Model for Simulating New Product Diffusion

The Bass model is one of the basic models for new product diffusion analysis. But the Bass model requires a large quantity of raw data to determine parameters of Bass model and this model also uses the potential capacity of market based on the subjective experience. To solve the problem of necessity of large raw data of Bass model, Wang put forward the Grey Bass model. Wang used non-linear least square (NLS) method to find the parameters of Grey Bass model and to assess the potential capacity of market. In the present paper a more appropriate method for Grey Bass equation is offered which estimates potential capacity of market even if the sample size is small. The proposed model is based on the minimization of sum of square of error between actual and predicted data using Particle Swarm Optimization (PSO) technique. Using the case study data, as used by Wang, the accuracy of the improved method is investigated. The results show that the mean absolute percentage error (MAPE) in the present case is 6.52 % compared to 7.93% reported by Wang.

InfraRed Astronomy Satellite Swarm Interferometry (IRASSI): Overview and study results

The far-infrared (FIR) regime is one of the few wavelength ranges where no astronomical data with sub-arcsecond spatial resolution exist yet. Neither of the medium-term satellite projects like SPICA, Millimetron or OST will resolve this malady. For many research areas, however, information at high spatial and spectral resolution in the FIR, taken from atomic fine-structure lines, from highly excited carbon monoxide (CO) and especially from water lines would open the door for transformative science. These demands call for interferometric concepts. We present here first results of our feasibility study IRASSI (Infrared Astronomy Satellite Swarm Interferometry) for an FIR space interferometer. Extending on the principal concept of the previous study ESPRIT, it features heterodyne interferometry within a swarm of five satellite elements. The satellites can drift in and out within a range of several hundred meters, thereby achieving spatial resolutions of <0.1 arcsec over the whole wavelength range of 1–6 THz. Precise knowledge on the baselines will be ensured by metrology methods employing laser-based optical frequency combs, for which preliminary ground-based tests have been designed by members of our study team. We first give a motivation on how the science requirements translate into operational and design parameters for IRASSI. Our consortium has put much emphasis on the navigational aspects of such a free-flying swarm of satellites operating in relatively close vicinity. We hence present work on the formation geometry, the relative dynamics of the swarm, and aspects of our investigation towards attitude estimation. Furthermore, we discuss issues regarding the real-time capability of the autonomous relative positioning system, which is an important aspect for IRASSI where, due to the large raw data rates expected, the interferometric correlation has to be done onboard, in quasi-real-time. We also address questions regarding the spacecraft architecture and how a first thermomechanical model is used to study the effect of thermal perturbations on the spacecraft. This will have implications for the necessary internal calibration of the local tie between the laser metrology and the phase centres of the science signals and will ultimately affect the accuracy of the baseline estimations.

Sub-pixel image registration on an embedded Nanosatellite Platform

CubeSats are limited in their physical size, which limits the physical size of payloads they can carry, thereby limiting the image quality of CubeSat Earth Observation missions. Algorithms exist that combine partially overlapping images to produce better output image quality. These algorithms may either improve the signal-to-noise ratio via averaging, increase resolution via super-resolution or merely remove redundant information via mosaicing. Such algorithms typically only function properly if the geometric transformations between consecutive images are known with high accuracy. These algorithms can either be applied terrestrially or on-board a satellite. Downloading large raw image data sets for terrestrial processing is impractical for a CubeSat mission, and therefore an on-board solution is desirable.This paper discusses how to accurately determine the transformation between consecutive images on-board, laying the foundation for efficient on-board de-noising, super-resolution and mosaicing. Two common methods used to determine translation – normalised cross correlation (NCC) and phase correlation – are investigated. From simulated results, NCC is shown to be the better candidate for our application. NCC achieves sub-pixel accuracy by making use of polynomial least squares regression. NCC is well suited for implementation on a satellite platform where images are captured in quick succession, resulting in partially overlapping images with little rotation between frames.We compare two potential hardware platforms – the MicroZed 7020 and Jetson TK1 – and then describe how we implemented our proposed solution onto the former, using a hardware description language. The effect of tuning NCC's parameters – specifically window and template size – is investigated with regards to hardware utilisation and accuracy. Software simulation and firmware-implementation results, using simulated data, are compared and discussed. We conclude by discussing expected full system runtime and energy efficiency.

Self-reported and routinely collected electronic healthcare resource-use data for trial-based economic evaluations: the current state of play in England and considerations for the future

BackgroundRandomised controlled trials (RCTs) are generally regarded as the “gold standard” for providing quantifiable evidence around the effectiveness and cost-effectiveness of new healthcare technologies. In order to perform the economic evaluations associated with RCTs, there is a need for accessible and good quality resource-use data; for the purpose of discussion here, data that best reflect the care received.Traditionally, researchers have developed questionnaires for resource-use data collection. However, the evolution of routinely collected electronic data within care services provides new opportunities for collecting data without burdening patients or caregivers (e.g. clinicians). This paper describes the potential strengths and limitations of each data collection method and then discusses aspects for consideration before choosing which method to use.Main textWe describe electronic data sources (large observational datasets, commissioning data, and raw data extraction) that may be suitable data sources for informing clinical trials and the current status of self-reported instruments for measuring resource-use. We assess the methodological risks and benefits, and compare the two methodologies. We focus on healthcare resource-use; however, many of the considerations have relevance to clinical questions.Patient self-report forms a pragmatic and cheap method that is largely under the control of the researcher. However, there are known issues with the validity of the data collected, loss to follow-up may be high, and questionnaires suffer from missing data. Routinely collected electronic data may be more accurate and more practical if large numbers of patients are involved. However, datasets often incur a cost and researchers are bound by the time for data approval and extraction by the data holders.ConclusionsOwing to the issues associated with electronic datasets, self-reported methods may currently be the preferred option. However, electronic hospital data are relatively more accessible, informative, standardised, and reliable. Therefore in trials where secondary care constitutes a major driver of patient care, detailed electronic data may be considered superior to self-reported methods; with the caveat of requiring data sharing agreements with third party providers and potentially time-consuming extraction periods. Self-reported methods will still be required when a ‘societal’ perspective (e.g. quantifying informal care) is desirable for the intended analysis.

Tourism Trend Mapping Based on Social Media Using SAW Algorithm

Indonesia is a country that has a variety of places to hold holidays. Having a plan about the most popular tourist attractions is the key for everyone who wants to travel to an area. But data on tourist spots that are currently in trend are often not easily obtained. Different from tour agents who already have a lot of experience and data without fear of missing information. therefore, presenting the tourism trends in a particular group is almost impossible to use data on the number of frequented places. This paper will discuss tourism trends that have appeal based on social media, with case studies on Twitter data with certain regional fields. Social media provides large raw data including travelers’ choices when they visit the venue. however, the data obtained is not shaped in human language. These results will be processed first and will be arranged to determine the level of tourist destinations are often visited. The algorithm used to process using Simple Additive Weighting (SAW) is proposed to arrange food ranking. To measure accuracy, we compared our results with tourist sights in Yogyakarta.

Comparison of lossless compression schemes for high rate electrical grid time series for smart grid monitoring and analysis

The smart power grid of the future will utilize waveform level monitoring with sampling rates in the kilohertz range for detailed grid status assessment. To this end, we address the challenge of handling large raw data amount with its quasi-periodical characteristic via lossless compression. We compare different freely available algorithms and implementations with regard to compression ratio, computation time and working principle to find the most suitable compression strategy for this type of data. Algorithms from the audio domain (ALAC, ALS, APE, FLAC & TrueAudio) and general archiving schemes (LZMA, Delfate, PPMd, BZip2 & Gzip) are tested against each other. We assemble a dataset from openly available sources (UK-DALE, MIT-REDD, EDR) and establish dataset independent comparison criteria. This combination is a first detailed open benchmark to support the development of tailored lossless compression schemes and a decision support for researchers facing data intensive smart grid measurements.

ANALISIS FAKTOR YANG MEMENGARUHI AGLOMERASI INDUSTRI UNGGULAN DAERAH DAN HUBUNGANNYA DENGAN DAYA SAING INDUSTRI DAERAH

The manufacturing industry sector is a major driver of economic growth in Indonesia with the largest contribution to the components of Gross Domestic Product is 25.60 % in 2012. But the globalization and liberalization of international trade requires industries to be more competitive. Improving the competitiveness of the industry can be done through the development of regional-based industrial sector main industry that area. The purpose of this study was to analyze the competitiveness and industrial agglomeration, the causality relationship between competitiveness and agglomeration industry and the factors that influence agglomeration of province main industries. This study was using large and medium scale industry raw data. The data analysis using Location Quotient, Hoover Balassa Index, Granger Causality method and panel data method with Fixed Effect Model. The result of panel data regression shows factors that influence the agglomeration of province main industries are firm size, value added, the diversity of industry, industry competition index, competitiveness index, wages and road infastructure. Keywords: Agglomeration, Granger Causality, Industrial Competitiveness, Panel Data

ANALISIS FAKTOR YANG MEMENGARUHI AGLOMERASI INDUSTRI UNGGULAN DAERAH DAN HUBUNGANNYA DENGAN DAYA SAING INDUSTRI DAERAH

The manufacturing industry sector is a major driver of economic growth in Indonesia with the largest contribution to the components of Gross Domestic Product is 25.60 % in 2012. But the globalization and liberalization of international trade requires industries to be more competitive. Improving the competitiveness of the industry can be done through the development of regional-based industrial sector main industry that area. The purpose of this study was to analyze the competitiveness and industrial agglomeration, the causality relationship between competitiveness and agglomeration industry and the factors that influence agglomeration of province main industries. This study was using large and medium scale industry raw data. The data analysis using Location Quotient, Hoover Balassa Index, Granger Causality method and panel data method with Fixed Effect Model. The result of panel data regression shows factors that influence the agglomeration of province main industries are firm size, value added, the diversity of industry, industry competition index, competitiveness index, wages and road infastructure. Keywords: Agglomeration, Granger Causality, Industrial Competitiveness, Panel Data

Detection of formaldehyde oxidation catalysis by MCR-ALS analysis of multiset ToF-SIMS data in positive and negative modes

Toxicity of formaldehyde is extremely high even at very low concentration in air. We have used ToF-SIMS (Time of Flight Secondary Ion Mass Spectroscopy) to explore formaldehyde oxidation catalyzed by birnessite (Kx (Mn4+,Mn3+)2O4) in the temperature range from 25 to 200 °C. ToF-SIMS is a powerful tool for the generation of chemical maps because of its high spatial resolution, its sensitivity and relatively low acquisition time. In this method the mass analysis of positive and negative ions sputtered out of the uppermost layers of the analyzed sample is performed. ToF-SIMS produces large raw data sets with rich chemical information but rather complex to be analyzed and interpreted. This is why the application of chemometric methods is proposed in this work for the exploration of ToF-SIMS complex data sets. In this work, MCR-ALS (Multivariate Curve Resolution-Alternating Least Squares) has been applied to resolve both the positive and negative ions present in ToF-SIMS data sets simultaneously analyzed using a data matrix augmentation strategy. Birnessite without catalyzed formaldehyde was first analyzed to resolve background contributions and use them to implement a selectivity constraint for MCR-ALS to remove them in the analysis of the formaldehyde oxidation data sets. Results show that, following the temperature increase, concentration of formaldehyde combined with manganese ions decreased whereas concentration of manganese oxide increased. Conformation changes of manganese formaldehyde metal complex were then inferred. It is concluded that the formation of the metal complex species formed between two manganese ions and only one formaldehyde molecule is very unlikely to exist.