Unstructured Datasets Research Articles

Towards a Lifecycle Information Framework and Technology in Manufacturing.

Industry has been chasing the dream of integrating and linking data across the product lifecycle and enterprises for decades. However, industry has been challenged by the fact that the context in which data is used varies based on the function / role in the product lifecycle that is interacting with the data. Holistically, the data across the product lifecycle must be considered an unstructured data-set because multiple data repositories and domain-specific schema exist in each phase of the lifecycle. This paper explores a concept called the Lifecycle Information Framework and Technology (LIFT). LIFT is a conceptual framework for lifecycle information management and the integration of emerging and existing technologies, which together form the basis of a research agenda for dynamic information modeling in support of digital-data curation and reuse in manufacturing. This paper provides a discussion of the existing technologies and activities that the LIFT concept leverages. Also, the paper describes the motivation for applying such work to the domain of manufacturing. Then, the LIFT concept is discussed in detail, while underlying technologies are further examined and a use case is detailed. Lastly, potential impacts are explored.

A Way How to Impart Data Science Skills to Computer Science Students Exemplified by OBDA-Systems Development

Nowadays to explore and examine data from a variety of angles to tackle Big Data problems, to devise data-driven solutions to the most pressing challenges, it is necessary to build multidisciplinary students’ skills set for innovative methods not only for master’s in Data Science programs but in traditional Computer Science programs also. In the paper, we describe how teaching methods and tools, which are used to train students to develop Ontology-Based Data Access systems with natural language interface to relational databases helps Master’s Degree students in Computer Science to collaborate with Data Scientists in real-world interdisciplinary projects and to prepare them for a data science career. We use ontology engineering in a combination with Natural Language Processing methods based on lexico-syntactic patterns, in particular, to extract needful data from structured, semi-structured and unstructured datasets in a uniform way to analyze real-world Russian social networks related to new building area.

Storage-Tag-Aware Scheduler for Hadoop Cluster

Big data analytics has simplified the processing complexity of extremely large data sets through ecosystems, such as Hadoop, MapR, and Cloudera. Apache Hadoop is an open-source ecosystem that manages large data sets in a distributed environment. MapReduce is a programming model that processes massive amount of unstructured data sets over Hadoop cluster. Recently, Hadoop enhances its homogeneous storage function to heterogeneous storage and stores data sets into multiple storage media, i.e., SSD, RAM, and DISK. This development increases the performance of data block placement strategy and allows a client to store large data sets into multiple storage media efficiently than homogeneous storage. However, this evolution increases the consumption of computing capacity and memory usage over MapReduce job scheduling. The scheduler processes MapReduce job into homogeneous container having configuration of CPU, memory, DISK volume, and network I/O, and accesses, processes, and stores data sets over heterogeneous storage media. This produces a processing latency of locating and pairing source data set to MapReduce tasks and results an abnormal high consumption of computing capacity and memory usage in a Datanode. Similarly, when scheduler assigns MapReduce jobs to multiple Datanodes, the same processing latency can severely affect the performance of whole cluster. In this paper, we present Storage-Tag-Aware Scheduler (STAS) that reduces processing latency by scheduling MapReduce jobs into heterogeneous storage containers, i.e., SSD, DISK, and RAM container. STAS endorses job with a tag of storage media, such as $Job_{SSD}$ , $Job_{DISK}$ , and $Job_{RAM}$ and parses them into heterogeneous shared-queues, which assign processing configuration to enlist jobs. STAS manager then schedules shared-queue jobs into heterogeneous MapReduce containers and generates an output into storage media of the cluster. The experimental evaluation shows that STAS optimizes the consumption of computing capacity and memory usage efficiently than available schedulers in a Hadoop cluster.

Evaluation of Filtering Methods Applied to the Unstructured Datasets in the Predictive Learning Services

Predictive learning services perform aggregation and homogenization of open data from public sources, in particular from the online recruitment agencies. However, the sample of vacancies may contain various percentage of noise due to the frequent occurrence of homonyms. This article will consider two approaches of noise reduction: the first one is based on the cosine similarity and the second one is based on the contextual words.

IPB -Implementation of Parallel Mining for Big Data

Data in this era is generating at tremendous rate so now it is need of today to handle the data to gain useful insight, this data can be useful for researcher and accommodation to do analysis. As we know traditional system cannot handle more than terabytes of data since it affects performance and also storage is very costly. Big data is an innovative technique analyze, store, manage, distributes and capture datasets. Objective: To achieve compressed storage, we implement a parallel mining algorithm called as Implementation of Parallel Mining for Big data. Method: Hadoop is a platform which enables the distributing processing using map reduces programming. This help in getting result at very fast rate as result in less time help in competing for growth of business. Unstructured datasets is taken for analysis which is real time is taken and converted to structured format and process in map reduces. It is found in literature existing mining algorithm for the real time datasets which always lacks in fault tolerance, load balancing, data distribution and automatic parallelization. To overcome these disadvantages we implement map reduce for association analysis. Finding/Improvement: In IPB we improve performance in the computing node the load is distributed. In our proposed solution we use real-world celestial spectral data. The graphical representation of traditional system comparison with Hadoop is shown in this paper.

R&D Perspective Social Issue Packaging using Text Analysis

In recent years, text mining has been used to extract meaningful insights from the large volume of unstructured text data sets of various domains. As...

A unified framework for harmonic analysis of functions on directed graphs and changing data

We present a general framework for studying harmonic analysis of functions in the settings of various emerging problems in the theory of diffusion geometry. The starting point of the now classical diffusion geometry approach is the construction of a kernel whose discretization leads to an undirected graph structure on an unstructured data set. We study the question of constructing such kernels for directed graph structures, and argue that our construction is essentially the only way to do so using discretizations of kernels. We then use our previous theory to develop harmonic analysis based on the singular value decomposition of the resulting non-self-adjoint operators associated with the directed graph. Next, we consider the question of how functions defined on one space evolves to another space in the paradigm of changing data sets recently introduced by Coifman and Hirn. While the approach of Coifman and Hirn require that the points on one space should be in a known one-to-one correspondence with the points on the other, our approach allows the identification of only a subset of landmark points. We introduce a new definition of distance between points on two spaces, construct localized kernels based on the two spaces and certain interaction parameters, and study the evolution of smoothness of a function on one space to its lifting to the other space via the landmarks. We develop novel mathematical tools that enable us to study these seemingly different problems in a unified manner.

Experiences in integrated data and research object publishing using GigaDB

In the era of computation and data-driven research, traditional methods of disseminating research are no longer fit-for-purpose. New approaches for disseminating data, methods and results are required to maximize knowledge discovery. The “long tail” of small, unstructured datasets is well catered for by a number of general-purpose repositories, but there has been less support for “big data”. Outlined here are our experiences in attempting to tackle the gaps in publishing large-scale, computationally intensive research. GigaScience is an open-access, open-data journal aiming to revolutionize large-scale biological data dissemination, organization and re-use. Through use of the data handling infrastructure of the genomics centre BGI, GigaScience links standard manuscript publication with an integrated database (GigaDB) that hosts all associated data, and provides additional data analysis tools and computing resources. Furthermore, the supporting workflows and methods are also integrated to make published articles more transparent and open. GigaDB has released many new and previously unpublished datasets and data types, including as urgently needed data to tackle infectious disease outbreaks, cancer and the growing food crisis. Other “executable” research objects, such as workflows, virtual machines and software from several GigaScience articles have been archived and shared in reproducible, transparent and usable formats. With data citation producing evidence of, and credit for, its use in the wider research community, GigaScience demonstrates a move towards more executable publications. Here data analyses can be reproduced and built upon by users without coding backgrounds or heavy computational infrastructure in a more democratized manner.

Private Communication in Public Spaces: The Paradoxical Economics of Exceptional Access and E-Privacy

Recent regulatory developments have seemed to be heading in sharply divergent directions. Some (e.g. the UK’s Investigatory Powers Bill) expand government power to collect, retain, process and share data. Others, like the EU’s General Data Privacy Regulation (GDPR), the EU-US Privacy Shield and the Freedom USA Act (to a degree) protect individuals against new intrusions by public and private parties. The cybersecurity issues are by now well-known (e.g. Abelson et al. (2015) “Keys under doormats: mandating insecurity by requiring government access to all data and communications.” Journal of Cybersecurity and Zittrain et al. (2016) “Don't Panic: Making Progress on the ‘Going Dark’ Debate.” Berkman Center Research Publication 2016-1), as are the privacy dimensions and the unfortunate processes by which well-publicised incidents drive beliefs and trigger slow and difficult to reverse rulemaking. At the same time, the economic importance of data have been growing and the variety of technologies, business models market structures and activities through which data are generated and used have all been evolving. Despite this, the economic analysis of the mechanisms and effects of such rules remains rudimentary, at least in the policy process. This paper seeks to understand how the economics of this new landscape shapes this development and how these cases can influence modern economic thinking (e.g. about complex networks, behaviour and platforms). It analyses the economic mechanisms and likely effects of such rules, in isolation and together. It is based on the published impact assessments, economic modelling of multi-sided platforms and information security and on the economic analysis of cybersecurity and ethical issues connected with data handling. To illustrate the issue, note that much of the existing literature and policy thinking conflates communication with the flow of information, privacy with data privacy, and security with an attacker-defender struggle over specific information assets. Few of these simplifications are accurate; most lead to unintended consequences. Consider communication; much of the traffic flowing through electronic networks, retained in data repositories and processed to support automated decisions is not communication that seeks to convey messages between people. It may be no more than traces of one-sided personal activity or the collective operation of large automated systems. Even consciously sent messages may be addressed to audiences that are diffuse, remote in time or space or purely imaginary. But once observed or captured as data, they can be regarded as communications. This can have paradoxical effects. An individual’s playful or careless expression or experimental search for variety, or observable patterns in large unstructured data sets may be regarded as an expression of fixed opinions, intentions or preferences or system laws. Rules based on communication miss this and the importance of actors at other levels of the system, from hardware to services. Vertical interactions and market power are hard to identify, and harder still to take into account when making policy. And as we know from net neutrality, the different of these actors (a bit is a bit, a bitcoin is a bitcoin) lead to novel phenomena. The paper gives a critical analysis of existing scholarly and policy debates, noting the disagreements between e.g. security and privacy or commerce and policy. It then analyses four cases to show how economic impacts have been neglected and could be incorporated.

A big data MapReduce framework for fault diagnosis in cloud-based manufacturing

This research develops a MapReduce framework for automatic pattern recognition based on fault diagnosis by solving data imbalance problem in a cloud-based manufacturing (CBM). Fault diagnosis in a CBM system significantly contributes to reduce the product testing cost and enhances manufacturing quality. One of the major challenges facing the big data analytics in CBM is handling of data-sets, which are highly imbalanced in nature due to poor classification result when machine learning techniques are applied on such data-sets. The framework proposed in this research uses a hybrid approach to deal with big data-set for smarter decisions. Furthermore, we compare the performance of radial basis function-based Support Vector Machine classifier with standard techniques. Our findings suggest that the most important task in CBM is to predict the effect of data errors on quality due to highly imbalance unstructured data-set. The proposed framework is an original contribution to the body of literature, where our proposed MapReduce framework has been used for fault detection by managing data imbalance problem appropriately and relating it to firm’s profit function. The experimental results are validated using a case study of steel plate manufacturing fault diagnosis, with crucial performance matrices such as accuracy, specificity and sensitivity. A comparative study shows that the methods used in the proposed framework outperform the traditional ones.

Study of Data Mining Algorithms in the Context of Performance Enhancement of Classification

Data mining can help researchers to gain novel and deep insights for understanding of large datasets. Nowadays, people are using data mining algorithms in different contexts like banking, hospitals, marketing etc. Classification algorithm plays a vital role. In the study, we found that SVM is the best classifier amongst all the classifiers. Here we used learning algorithms with the historical dataset to train the classifier and the test samples are used to validate the correctness of the classifier. We might have structured semistructured and unstructured datasets which are used for classification. We have performed the study of reputed literatures that belong to classification area to identify some new enhancements in the classifiers. A few most important classifiers are SVM, decision tree, neural network, Naive Bayes. We found most of the literatures were concentrated on SVM classifier so we targeted SVM classifier for the performance enhancement. SVM are important tool in datamining to classify data. The aim of this review is to identify the effectiveness of kernel parameters for classification of data using Support Vector.

Big Folklore: A Special Issue on Computational Folkloristics

Any fact becomes important when it's connected to another.-Umberto Eco (1989:385)OVER THE COURSE OF THE PAST DECADE, a revolution has occurred in the materials available for the study of folklore. The size and scope of digital archives of folklore have exploded, and the magnitude of digital materials available for folkloristic consideration has increased exponentially. Around the world, national archives have made considerable efforts to make their resources machine-readable, while other initiatives have focused on the digitization of resources related to smaller regions, single collectors, or single genres. Simultaneously, the explosive growth in social media, weblogs (blogs), and other internet resources have made previously hard-to-access records of traditional expressive culture accessible at a scale so enormous that it is hard to fathom. These developments, coupled with the development of algorithmic approaches to the analysis of large, unstructured datasets and new methods for the visualization and navigation of the relationships discovered by these algorithmic approaches-from mapping to 3-D embedding, from time lines to navigable visualizations-offer folklorists new opportunities for the analysis of traditional expression. Folklore studies that leverage the power of these algorithmic approaches fall under the rubric of computational folkloristics (Abello, broadwell, and tangherlini 2012).Certain challenges attach to work in the digital realm, particularly if one believes that folklore emerges from the productive dialectic that exists between individuals and tradition. From this perspective, the study of folklore is predicated on the retrospective analysis of recordings of culturally expressive forms and their performance or articulation. in earlier work, i outlined four main areas that will, in the coming years, require significant attention as the materials of study are increasingly represented as digital objects, whether the materials be oral performances or aspects of material culture, or any of the other numerous types of expression that folklorists work with (tangherlini 2013c). broadly speaking, these four areas are (1) collecting and archiving, (2) indexing and classifying, (3) visualization and navigation, and (4) analysis (tangherlini 2013c:8).Collecting and ArchivingPrior to the advent of the digital age, fieldwork was an individual or small group endeavor that frequently generated idiosyncratic but simple data that could be easily incorporated into the analog archives of either the researcher or larger institutions. by way of contrast, contemporary digitally based fieldwork often generates huge amounts of data in formats that are not always stable or easily accessed by future researchers.1 inevitably, efforts to transform archival materials from mostly analog formats that inadvertently regulate access (one must travel to use them) to digital formats that can theoretically be made freely available throughout the world, raise complicated intellectual property issues. Although these issues lie outside the realm of the computational, they necessarily affect how folklorists work.2The World Wide Web (www) has also led to the development of new collecting methodologies. Whereas most folklore fieldwork in the past was carried out either through face-to-face interaction or surveys, fieldwork can now be carried out on and among (as opposed to with) groups and individuals who are not necessarily aware that they are participating in an ethnographic project. Predicated on the conceptualization of the Web as a dynamic, self-organizing folklore archive, these efforts rely on several algorithmic approaches to data collection, from simple scraping to Web crawling.3 one can, for example, fairly easily create a spider that crawls the net, storing not only content related to the parameters of the crawl, but also metadata about the sites visited and the users of those sites.4 consequently, these collection methods precipitate profound ethical concerns related to privacy and surveillance. …

Cross-domain based Event Recommendation using Tensor Factorization

Abstract Context in the form of meta-data has been accredited as an important component in cross-domain collaborative filtering (CDCF). In this research paper CDCF concept is used to exploit event information (context) from two UI matrices to allow the recommendation performance of one domain (Facebook- User-Event Matrix) to benefit from the information from another domain (Bookmyshow- Event-Tag Matrix). The model based collaborative filtering approach Tensor Factorization(TF) has been used to integrate Facebook provided User-Event context information with Bookmyshow Event-Tag context information to recommend events. In contrast to the standard collaborative tag recommendation, our CDCF approach uses one User-Event matrix of Facebook that takes another Bookmyshow Event-Tag matrix as additional informant. The proposed cross-domain based Event Recommendation approach is divided into three modules- i) data collection which extracts the unstructured dataset from the two domains Bookmyshow and social networking site Facebook using API’s; ii) data mapping module which is basically used to integrate the common knowledge/ data that can be shared between considered different domains (Facebook & Bookmyshow). This module integrates and reduces the data into structured events’ instances. As the dataset was collected from two different sites, an intersection of both was taken out. Therefore this module is carefully designed according to reliability of information that is common between two domains; iii) 3 order tensor factorization and Latent Dirichlet Allocation (LDA) used for most preferable recommendation by less pertinent result reduction. The proposed 3 order tensor factorization is designed for maximizing the mutual benefit from both the considered domains (organizer and user). Therefore providing three recommendations: For organizers: 1) system recommends places to conduct specific event according to maximum of attendees of a particular type of event at a specific location; 2) recommending target audience to organizer: those who are interested to attend event on the basis of past data for promotion purposes. For users: 3) recommending events to users of their interest on the basis of past record. Our result shows significant improvement in reduction of less relevant data and result effectiveness is measured through recall and precision. Reduction of less relevant recommendation is 64%, 72% and 63% for place recommendation to organizer, target audience recommendation to organizer and event recommendation to user respectively. The proposed tensor factorization approach achieved 68% precision, 15.5% recall in recommending attendees to organizer and 62% precision, 13.4% recall for event recommendation to user.

A hybrid spam detection method based on unstructured datasets

The identification of non-genuine or malicious messages poses a variety of challenges due to the continuous changes in the techniques utilised by cyber-criminals. In this article, we propose a hybrid detection method based on a combination of image and text spam recognition techniques. In particular, the former is based on sparse representation-based classification, which focuses on the global and local image features, and a dictionary learning technique to achieve a spam and a ham sub-dictionary. On the other hand, the textual analysis is based on semantic properties of documents to assess the level of maliciousness. More specifically, we are able to distinguish between meta-spam and real spam. Experimental results show the accuracy and potential of our approach.

Datafication and the Seductive Power of Uncertainty—A Critical Exploration of Big Data Enthusiasm

This contribution explores the fine line between overestimated expectations and underrepresented momentums of uncertainty that correlate with the prevalence of big data. Big data promises a multitude of innovative options to enhance decision-making by employing algorithmic power to gather worthy information out of large unstructured data sets. Datafication—the exploitation of raw data in many different contexts—can be seen as an attempt to tackle complexity and reduce uncertainty. Accordingly promising are the prospects for innovative applications to gain new insights and valuable knowledge in a variety of domains ranging from business strategy, security to health and medical research, etc. However, big data also entails an increase in complexity that, together with growing automation, may trigger not merely uncertain but also unintended societal events. As a new source of networking power, big data has inherent risks to create new asymmetries and transform possibilities to probabilities that can inter alia affect the autonomy of the individual. To reduce these risks, challenges ahead include improving data quality and interpretation supported by new modalities to allow for scrutiny and verifiability of big data analytics.

K-Means Clustering For Segment Web Search Results

Clustering is the power full technique for segment relevant data into different levels. This study has proposed K- means clustering method for cluster web search results for search engines. For represent documents we used vector space model and use cosine similarity method for measure similarity between user query and the search results. As an improvement of K-means clustering we used distortion curve method for identify optimal initial number of clusters. should exhibit a large degree of similarity while the similarity among different clusters should be minimized. Some of the more familiar clustering methods are: partitioning algorithms based on dividing entire data into dissimilar groups, hierarchical methods, density and grid based clustering, some graph based methods, etc (3). Most popular and widely use partitioning method is K-means clustering method. A. K-means Clustering K-Means Clustering is a method used to classify semi structured or unstructured data sets. This is one of the most commonly and effective methods to classify data because of its simplicity and ability to handle voluminous data sets. It accepts the number of clusters and the initial set of centroids as parameters. The distance of each item in the data set is calculated with each of the centroids of the respective cluster. The item is then assigned to the cluster with which the distance of the item is the least. The centroid of the cluster to which the item was assigned is recalculated. One of the most important and commonly used methods for grouping the items of a data set using K-Means Clustering is calculating the distance of the point from the chosen mean. This distance is usually the Euclidean Distance. ∑ √( )

May the Sorts Be with You

More and more investors are tracking and analyzing news flows and text sentiments in large, unstructured datasets and mining data in a split second for investment-relevant information. Often, the goal is to incorporate findings into trading strategies and algorithms, to back-test investment theses, or to improve risk-monitoring activities. Finding information of interest in large, textual databases is becoming easier as semantic analytic technologies mature. Tools already abound to help companies monitor the sentiment toward their brands by evaluating “big data” repositories, such as customer comments. Text and news analytics for investing is a newer niche application, and many are now experimenting with investment use cases. As they do, the need to quantify and incorporate unstructured data into investing equations continues to grow, along with the volume of textual data: About 2.5 billion gigabytes of data were generated every day in 2012, according to IBM, and by some estimates, 75% of data are unstructured. Data sources that drive tools for investors are diverse and depend on the vendor of the analytics. Some vendors allow users to hand-pick data and custom aggregate it. A few examples of data sources for text analytics are news feeds, select blogs, social media data, RSS (rich site summary) feeds, customer relationship management (CRM) data, and audio and TV news transcripts, such as CNBC’s Squawk Box transcripts. News analytics have been used for years in equity investing, but now, investors are applying these tools in global macro strategies and foreign exchange (FX) trading as well as in commodities and fixed-income investing. Early adopters include quantitative hedge funds in the United States and United Kingdom. Machine learning capabilities are getting better at interpreting and almost understanding what news may mean, although many pitfalls still exist. Most news analytics tools use sentiment analysis algorithms that work with a simple scoring mechanism. Using natural language processing, the algorithm counts the number of positive words and the number of negative words in a text and then assesses which company name appears most in the text. The algorithm assigns a positive or negative score to the text, often including a relevance or probability score for the different entities. It then aggregates this data over the whole news flow and gives users an indication of whether certain companies or terms are trending positive or negative. Some tools score news sentiment and present it as a time series, measure the volume of media coverage, or capture deviations in media coverage that can be telling for investors. Other tools define and track entity relationships in massive textual datasets. The focus of news analytics tools is no longer on speed or reducing latency, according to Armando Gonzalez, president and CEO of RavenPack, a provider of news analytics for quantitative finance applications. “It’s about developing sustainable alpha or verifiable risk management strategies from unstructured data,” he says. “Can users reliably extract a durable signal from the vast amount of news and social content out there that lives for hours, days, weeks, and ideally months?”

Reduced Topologically Real-World Networks

The topological and dynamical properties of real-world networks have attracted extensive research from a variety of multi-disciplinary fields. They, in fact, model typically big datasets which pose interesting challenges, due to their intrinsic size and complex interactions, as well as the dependencies between their different sub-parts. Therefore, defining networks based on such properties, is unlikely to produce usable information due to their complexity and the data inconsistencies which they typically contain. In this paper, the authors discuss the evaluation of a method as part of ongoing research which aims to mine data to assess whether their associated networks exhibit properties comparable to well-known structures, namely scale-free, small world and random networks. For this, they will use a large dataset containing information on the seismologic activity recorded by the European-Mediterranean Seismological Centre. The authors will show that it provides an accurate, agile, and scalable tool to extract useful information. This further motivates their effort to produce a big data analytics tool which will focus on obtaining in-depth intelligence from both structured and unstructured big datasets. This will ultimately lead to a better understanding and prediction of the properties of the system(s) they model.

A survey on text mining in social networks

AbstractIn this survey, we review different text mining techniques to discover various textual patterns from the social networking sites. Social network applications create opportunities to establish interaction among people leading to mutual learning and sharing of valuable knowledge, such as chat, comments, and discussion boards. Data in social networking websites is inherently unstructured and fuzzy in nature. In everyday life conversations, people do not care about the spellings and accurate grammatical construction of a sentence that may lead to different types of ambiguities, such as lexical, syntactic, and semantic. Therefore, analyzing and extracting information patterns from such data sets are more complex. Several surveys have been conducted to analyze different methods for the information extraction. Most of the surveys emphasized on the application of different text mining techniques for unstructured data sets reside in the form of text documents, but do not specifically target the data sets in social networking website. This survey attempts to provide a thorough understanding of different text mining techniques as well as the application of these techniques in the social networking websites. This survey investigates the recent advancement in the field of text analysis and covers two basic approaches of text mining, such as classification and clustering that are widely used for the exploration of the unstructured text available on the Web.

Estimating the temperature evolution of foodstuffs during freezing with a 3D meshless numerical method

Freezing processes are characterised by sharp changes in specific heat capacity and thermal conductivity for temperatures close to the freezing point. This leads to strong nonlinearities in the governing PDE that may be difficult to resolve using traditional numerical methods. In this work we present a meshless numerical method, based on a local Hermite radial basis function collocation approach in finite differencing mode, to allow the solution of freezing problems. By introducing a Kirchhoff transformation and solving the governing equations in Kirchhoff space, the strength of nonlinearity is reduced while preserving the structure of the heat equation. In combination with the high-resolution meshless numerical method, this allows efficient and stable numerical solutions to be obtained for freezing problems using 3D unstructured datasets. We demonstrate the proposed numerical formulation for the freezing of foodstuffs. By approximating the shape of the thermal conductivity and heat capacity curves in a piecewise linear fashion the temperature-dependent material curves may be described using eight independent parameters. We consider the optimisation of these parameters to match experimental data for the freezing of a hemispherical sample of mashed potato by using a simple manual procedure that requires a minimal number of simulations to be performed. Working in this way, a good approximation is obtained to the temperature profile throughout the sample without introducing instability into the numerical results.