Automatic extraction of concepts from texts and applications

The extraction of relevant terms from texts is an extensively researched task in Text- Mining. Relevant terms have been applied in areas such as Information Retrieval or document clustering and classification. However, relevance has a rather fuzzy nature since the classification of some terms as relevant or not relevant is not consensual. For instance, while words such as "president" and "republic" are generally considered relevant by human evaluators, and words like "the" and "or" are not, terms such as "read" and "finish" gather no consensus about their semantic and informativeness. Concepts, on the other hand, have a less fuzzy nature. Therefore, instead of deciding on the relevance of a term during the extraction phase, as most extractors do, I propose to first extract, from texts, what I have called generic concepts (all concepts) and postpone the decision about relevance for downstream applications, accordingly to their needs. For instance, a keyword extractor may assume that the most relevant keywords are the most frequent concepts on the documents. Moreover, most statistical extractors are incapable of extracting single-word and multi-word expressions using the same methodology. These factors led to the development of the ConceptExtractor, a statistical and language-independent methodology which is explained in Part I of this thesis. In Part II, I will show that the automatic extraction of concepts has great applicability. For instance, for the extraction of keywords from documents, using the Tf-Idf metric only on concepts yields better results than using Tf-Idf without concepts, specially for multi-words. In addition, since concepts can be semantically related to other concepts, this allows us to build implicit document descriptors. These applications led to published work. Finally, I will present some work that, although not published yet, is briefly discussed in this document.

Numerical prediction of diffusion and electric field-induced iron nanoparticle transport

Zero valent iron nanoparticles (nZVI) are considered very promising for the remediation of contaminated soils and groundwaters. However, an important issue related to their limited mobility remains unsolved. Direct current can be used to enhance the nanoparticles transport, based on the same principles of electrokinetic remediation. In this work, a generalized physicochemical model was developed and solved numerically to describe the nZVI transport through porous media under electric field, and with different electrolytes (with different ionic strengths). The model consists of the Nernst–Planck coupled system of equations, which accounts for the mass balance of ionic species in a fluid medium, when both the diffusion and electromigration of the ions are considered. The diffusion and electrophoretic transport of the negatively charged nZVI particles were also considered in the system. The contribution of electroosmotic flow to the overall mass transport was included in the model for all cases. The nZVI effective mobility values in the porous medium are very low (10−7–10−4 cm2 V−1 s−1), due to the counterbalance between the positive electroosmotic flow and the electrophoretic transport of the negatively charged nanoparticles. The higher the nZVI concentration is in the matrix, the higher the aggregation; therefore, low concentration of nZVI suspensions must be used for successful field application.

Video metadata extraction in a videoMail system

Currently the world swiftly adapts to visual communication. Online services like YouTube and Vine show that video is no longer the domain of broadcast television only. Video is used for different purposes like entertainment, information, education or communication. The rapid growth of today’s video archives with sparsely available editorial data creates a big problem of its retrieval. The humans see a video like a complex interplay of cognitive concepts. As a result there is a need to build a bridge between numeric values and semantic concepts. This establishes a connection that will facilitate videos’ retrieval by humans. The critical aspect of this bridge is video annotation. The process could be done manually or automatically. Manual annotation is very tedious, subjective and expensive. Therefore automatic annotation is being actively studied. In this thesis we focus on the multimedia content automatic annotation. Namely the use of analysis techniques for information retrieval allowing to automatically extract metadata from video in a videomail system. Furthermore the identification of text, people, actions, spaces, objects, including animals and plants. Hence it will be possible to align multimedia content with the text presented in the email message and the creation of applications for semantic video database indexing and retrieving.

Antioxidant extraction process for Andean Oca by a photodiode array detector using response surface methodology

Application of Experimental Design techniques has proven to be essential in various research fields, due to its statistical capability of processing the effect of interactions among independent variables, known as factors, in a system’s response. Advantages of this methodology can be summarized in more resource and time efficient experimentations while providing more accurate results. This research emphasizes the quantification of 4 antioxidants extraction, at two different concentration, prepared according to an experimental procedure and measured by a Photodiode Array Detector. Experimental planning was made following a Central Composite Design, which is a type of DoE that allows to consider the quadratic component in Response Surfaces, a component that includes pure curvature studies on the model produced. This work was executed with the intention of analyzing responses, peak areas obtained from chromatograms plotted by the Detector’s system, and comprehending if the factors considered – acquired from an extensive literary review – produced the expected effect in response. Completion of this work will allow to take conclusions regarding what factors should be considered for the optimization studies of antioxidants extraction in a Oca (Oxalis tuberosa) matrix.

Valorization of grape pomace through hot compressed water extraction/hydrolysis

The world energy consumption is expected to increase strongly in coming years, because of the emerging economies. Biomass is the only renewable carbon resource that is abundant enough to be used as a source of energy Grape pomace is one of the most abundant agro-industrial residues in the world, being a good biomass resource. The aim of this work is the valorization of grape pomace from white grapes (WWGP) and from red grapes (RWGP), through the extraction of phenolic compounds with antioxidant activity, as well as through the extraction/hydrolysis of carbohydrates, using subcritical water, or hot compressed water (HCW). The main focus of this work is the optimization of the process for WWGP, while for RWGP only one set of parameters were tested. The temperatures used were 170, 190 and 210 °C for WWGP, and 180 °C for RWGP. The water flow rates were 5 and 10 mL/min, and the pressure was always kept at 100 bar. Before performing HCW assays, both residues were characterized, revealing that WWGP is very rich in free sugars (around 40%) essentially glucose and fructose, while RWGP has higher contents of structural sugars, lignin, lipids and protein. For WWGP the best results were achieved at 210 °C and 10 mL/min: higher yield in water soluble compounds (69 wt.%), phenolics extraction (26.2 mg/g) and carbohydrates recovery (49.3 wt.% relative to the existing 57.8%). For RWGP the conditions were not optimized (180 °C and 5 mL/min), and the values of the yield in water soluble compounds (25 wt.%), phenolics extraction (19.5 mg/g) and carbohydrates recovery (11.4 wt.% relative to the existing 33.5%) were much lower. The antioxidant activity of the HCW extracts from each assay was determined, the best result being obtained for WWGP, namely for extracts obtained at 210 °C (EC50=20.8 μg/mL; EC50 = half maximum effective concentration; EC50 = 22.1 μg/mL for RWGP, at 180 ºC).

Engineered MRI nanoprobes based on superparamagnetic iron oxide nanoparticles

This project aimed to engineer new T2 MRI contrast agents for cell labeling based on formulations containing monodisperse iron oxide magnetic nanoparticles (MNP) coated with natural and synthetic polymers. Monodisperse MNP capped with hydrophobic ligands were synthesized by a thermal decomposition method, and further stabilized in aqueous media with citric acid or meso-2,3-dimercaptosuccinic acid (DMSA) through a ligand exchange reaction. Hydrophilic MNP-DMSA, with optimal hydrodynamic size distribution, colloidal stability and magnetic properties, were used for further functionalization with different coating materials. A covalent coupling strategy was devised to bind the biopolymer gum Arabic (GA) onto MNPDMSA and produce an efficient contrast agent, which enhanced cellular uptake in human colorectal carcinoma cells (HCT116 cell line) compared to uncoated MNP-DMSA. A similar protocol was employed to coat MNP-DMSA with a novel biopolymer produced by a biotechnological process, the exopolysaccharide (EPS) Fucopol. Similar to MNP-DMSA-GA, MNP-DMSA-EPS improved cellular uptake in HCT116 cells compared to MNP-DMSA. However, MNP-DMSA-EPS were particularly efficient towards the neural stem/progenitor cell line ReNcell VM, for which a better iron dose-dependent MRI contrast enhancement was obtained at low iron concentrations and short incubation times. A combination of synthetic and biological coating materials was also explored in this project, to design a dynamic tumortargeting nanoprobe activated by the acidic pH of tumors. The pH-dependent affinity pair neutravidin/iminobiotin, was combined in a multilayer architecture with the synthetic polymers poy-L-lysine and poly(ethylene glycol) and yielded an efficient MRI nanoprobe with ability to distinguish cells cultured in acidic pH conditions form cells cultured in physiological pH conditions.

Semantic relations extraction from unstructured information for domain ontologies enrichment

Based in internet growth, through semantic web, together with communication speed improvement and fast development of storage device sizes, data and information volume rises considerably every day. Because of this, in the last few years there has been a growing interest in structures for formal representation with suitable characteristics, such as the possibility to organize data and information, as well as the reuse of its contents aimed for the generation of new knowledge. Controlled Vocabulary, specifically Ontologies, present themselves in the lead as one of such structures of representation with high potential. Not only allow for data representation, as well as the reuse of such data for knowledge extraction, coupled with its subsequent storage through not so complex formalisms. However, for the purpose of assuring that ontology knowledge is always up to date, they need maintenance. Ontology Learning is an area which studies the details of update and maintenance of ontologies. It is worth noting that relevant literature already presents first results on automatic maintenance of ontologies, but still in a very early stage. Human-based processes are still the current way to update and maintain an ontology, which turns this into a cumbersome task. The generation of new knowledge aimed for ontology growth can be done based in Data Mining techniques, which is an area that studies techniques for data processing, pattern discovery and knowledge extraction in IT systems. This work aims at proposing a novel semi-automatic method for knowledge extraction from unstructured data sources, using Data Mining techniques, namely through pattern discovery, focused in improving the precision of concept and its semantic relations present in an ontology. In order to verify the applicability of the proposed method, a proof of concept was developed, presenting its results, which were applied in building and construction sector.

An agent based architecture to support monitoring in plug and produce manufacturing systems using knowledge extraction

In recent years a set of production paradigms were proposed in order to capacitate manufacturers to meet the new market requirements, such as the shift in demand for highly customized products resulting in a shorter product life cycle, rather than the traditional mass production standardized consumables. These new paradigms advocate solutions capable of facing these requirements, empowering manufacturing systems with a high capacity to adapt along with elevated flexibility and robustness in order to deal with disturbances, like unexpected orders or malfunctions. Evolvable Production Systems propose a solution based on the usage of modularity and self-organization with a fine granularity level, supporting pluggability and in this way allowing companies to add and/or remove components during execution without any extra re-programming effort. However, current monitoring software was not designed to fully support these characteristics, being commonly based on centralized SCADA systems, incapable of re-adapting during execution to the unexpected plugging/unplugging of devices nor changes in the entire system’s topology. Considering these aspects, the work developed for this thesis encompasses a fully distributed agent-based architecture, capable of performing knowledge extraction at different levels of abstraction without sacrificing the capacity to add and/or remove monitoring entities, responsible for data extraction and analysis, during runtime.