Building a knowledge graph by using cross-lingual transfer method and distributed MinIE algorithm on apache spark

Derived from knowledge bases, knowledge graphs represent knowledge expressions in graphs, which utilize nodes and edges to denote entities and relations conceptually. Knowledge graph can be described in textual triple form, consisting of head entities, tail entities and relations between entities. In order to represent elements in knowledge graphs, knowledge graph embedding techniques are proposed to map entities and relations into continuous vector spaces as numeric vectors for computational efficiency. Convolution-based knowledge graph embedding models have promising performance for knowledge graph representation learning. However, the input of those neural network-based models is frequently in handmade forms and may suffer from low efficiency in feature extraction procedure of the models. In this paper, a convolutional autoencoder is proposed for knowledge graph representation learning with entity pairs as input, aiming to obtain corresponding hidden relation representation. In addition, a bi-directional relation encoding network is utilized to represent semantic of entities in different directional relation patterns, as an encoder to output representation for initialization of the convolutional autoencoder. Experiments are conducted on standard datasets including, WN18RR, Kinship, NELL-995 and FB15k-237 as a link prediction task. Besides, input embedding matrix composed of different ingredients is designed to evaluate performances of the convolutional autoencoder. The results demonstrate that our model is effective in learning representation from entity feature interactions.

With the advancement in technology in recent years, IT administration and its infrastructure work on vast amount of data that needs to be stored and managed properly, for instance, handling of microservices which contains numerous systems. Client Complaint Management System is one such application system in every company. The IT Customer Support Department handles the complaints/queries of the customers where managing existing databases for support team is difficult due to its limitations to manage and maintain relationship of increasingly complex data. Knowledge Graph is a graph Database which stores and manages complex data by maintaining nodes and edges which define relationship between the nodes. This paper proposes an idea to create Knowledge graph for handling client complaint management system’s data with intention to solve the customer queries quickly and efficiently by using Naive Bayesian Classifier and Artificial Neural Networks for Text Classification, Natural Language Processing techniques and tools for generation of Knowledge graph. Using Knowledge Graph in this System for storing data will handle complex data relationships, reduce cost, ease the access for IT Administrators and staff for cross-referencing, increasing the speed and efficiency of the Data Management techniques as well.

The research results presented in this paper were obtained as part of the D2KAB project (Data to Knowledge in Agriculture and Biodiversity) which aims to develop semantic web-based tools to describe and make agronomical data actionable and accessible following the FAIR principles. We focus on constructing domain-specific Knowledge Graphs (KGs) from textual data sources, using Natural Language Processing (NLP) techniques to extract and structure relevant entities. Our approach is based on the formalization of a semantic data model using common linked open vocabularies such as the Web Annotation Ontology (OA) and the Provenance Ontology (PROV). The model was developed by formulating motivating scenarios and competency questions from domain experts. This model has been used to construct three different KGs from three distinct corpora: PubMed scientific publications on wheat and rice genetics and phenotyping, and French agricultural alert bulletins. The named entities to be recognized include genes, phenotypes, traits, genetic markers, taxa and phenological stages normalized using semantic resources such as the Wheat Trait and Phenotype Ontology (WTO), the French Crop Usage (FCU) thesaurus and the Plant Phenological Description Ontology (PPDO). Named entities were extracted using different NLP approaches and tools. The relevance of the semantic model was validated by implementing experts questions as SPARQL queries to be answered on the constructed RDF knowledge graphs. Our work demonstrates how domain-specific vocabularies and systematic querying of KGs can reveal hidden interactions and support agronomists in navigating vast amounts of data. The resources and transformation pipelines developed are publicly available in Git repositories.

Introduction:Nowadays in Panama, there is a lot of patient information stored in textual form which cannot be manipulated to manage adequate knowledge. There are multiple resources created to represent knowledge, including specialized glossaries, ontologies, among others. The ontologies are an important part within the scope of the recovery and organization of the information and the semantic web. Also in recent works they are used in applications of natural language processing (NLP), as a knowledge base.Aim:This research was conducted with the aim of creating a methodology that allows from a text written in NL, extract the necessary elements using NLP tools and with them create a knowledge base represented by one domain ontology and extract knowledge to help medical specialists.Material and Methods:In this study we carried out a methodology that allows the extraction of knowledge of patient clinical records, general medicine and palliative care, in order to show relevant knowledge elements to specialists. The methodology was validated with a data corpus of approximately 200 patient records.Conclusion:We have created a knowledge representation methodology, combining NLP techniques and tools and the automatic instantiation of an ontology, which can serve as a software agent for other applications or used to visualize the patient’s clinical information. The study was validated using the traditional metrics of information retrieval systems precision, recall, F-measure obtaining excellent results, and can be used as a software agent or methodology for the development of information extraction software systems in the medical domain.

Over the past few years, Large Language Models of Code (Code LLMs) have started to have a significant impact on programming practice. Code LLMs are also emerging as building blocks for research in programming languages and software engineering. However, the quality of code produced by a Code LLM varies significantly by programming language. Code LLMs produce impressive results on high-resource programming languages that are well represented in their training data (e.g., Java, Python, or JavaScript), but struggle with low-resource languages that have limited training data available (e.g., OCaml, Racket, and several others). This paper presents an effective approach for boosting the performance of Code LLMs on low-resource languages using semi-synthetic data. Our approach, called M ulti PL-T, generates high-quality datasets for low-resource languages, which can then be used to fine-tune any pretrained Code LLM. M ulti PL-T translates training data from high-resource languages into training data for low-resource languages in the following way. 1) We use a Code LLM to synthesize unit tests for commented code from a high-resource source language, filtering out faulty tests and code with low test coverage. 2) We use a Code LLM to translate the code from the high-resource source language to a target low-resource language. This gives us a corpus of candidate training data in the target language, but many of these translations are wrong. 3) We use a lightweight compiler to compile the test cases generated in (1) from the source language to the target language, which allows us to filter our obviously wrong translations. The result is a training corpus in the target low-resource language where all items have been validated with test cases. We apply this approach to generate tens of thousands of new, validated training items for five low-resource languages: Julia, Lua, OCaml, R, and Racket, using Python as the source high-resource language. Furthermore, we use an open Code LLM (StarCoderBase) with open training data (The Stack), which allows us to decontaminate benchmarks, train models without violating licenses, and run experiments that could not otherwise be done. Using datasets generated with M ulti PL-T, we present fine-tuned versions of StarCoderBase and Code Llama for Julia, Lua, OCaml, R, and Racket that outperform other fine-tunes of these base models on the natural language to code task. We also present Racket fine-tunes for two very recent models, DeepSeek Coder and StarCoder2, to show that M ulti PL-T continues to outperform other fine-tuning approaches for low-resource languages. The M ulti PL-T approach is easy to apply to new languages, and is significantly more efficient and effective than alternatives such as training longer.

Integration of Spark framework in Supply Chain Management

The exponential increase in Medical data and Computer-aided diagnostic tools has made it easier for Machine Learning (ML) Researchers to extract valuable insights from medical images resulting in better patient outcomes during Medical Summarization. According to current statistics, 8.8 million people died every year due to cancer death. It is therefore imperative to use these available image datasets for better diagnosis, increased patient outcomes and recovery from deadly diseases. As a result, this study proposes a Colorectal Cancer classification model for detecting KRAS mutation status of Patients through their Radiomic images using the transfer learning of a deep learning pipeline on Apache Spark. The datasets will be curated from National Cancer Institute (NCI), features extraction or classification of Computed tomography(CT) images into CT- based handcrafted radiomic signatures with a Convolutional Neural Network(CNN) using the Apache Spark framework, the system uploads the segmented scanning images to the High Distributed File System (HDFS), Confidence Interval(CI), Area Under Curve(AUC), Precision and Recall for Validation cohort, Docker for app deployment and containerization of the model developed for reproducibility, transfer learning and model reuse. The result would be a real-time model using Deep learning pipelines with Apache Spark and Keras Tensor flow for KRAS Mutation detection using Colorectal Cancer images. In conclusion, this research project would produce a scalable and reproducible model for the faster diagnosis of Colorectal Cancer through the KRAS mutation status of CRC Patients. Citation Format: Mary Adetutu Adewunmi. Scalable colorectal cancer (CRC) diagnosis model with arterial and unenhanced phases of KRAS mutational status using Apache Spark [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 6391.

In this article, constructing a Sentiment-embedded Event Evolutionary Graph (SEEGraph) is proposed to capture the dynamic relationships and narratives embedded in the oral histories of diasporas. It provides data support for applying diaspora oral archives knowledge and realizes the effective transformation of “data-knowledge-application.” Firstly, diaspora oral archives are introduced as the research scope. Secondly, domain ontology is utilized to build a knowledge representation system for diaspora oral archives. Based on the ontology, events and event relations are defined to construct SEEGraph. Next, the South Asian Oral History Project (SAOHP) is selected as the data source. Events entities and relationships involved in the diaspora oral archives are extracted using Natural Language Processing (NLP) tools, which form the data layer of the event evolutionary map. SEEGraph is then constructed. This step stores and maps instances extracted from archival data, transforming data into knowledge. Finally, the knowledge graph is applied to analyze and explore the event evolutionary patterns, spatio-temporal and spatio-sentiment rules of historical events, and emotional changes of diasporas. This innovative methodology not only provides valuable insights into the experiences, identities, and connections of diaspora members but also offers a comprehensive framework for understanding the complex interplay between mobility and belonging in diaspora oral archives.

This report describes the adaptation, composition and use of natural language processing, machine learning and other computational tools to help make implicit informational structures in very large technical corpora explicit. The tools applied to the corpora automatically build normalized multi-word term structures, which in turn are used to build taxonomies, semantic schema, topic models and knowledge graphs. In our cybersecurity use case, we apply these tools to help us understand the threat landscape as exhibited in the Common Vulnerabilities and Exposures (CVE - https://cve.mitre.org/ and https://nvd.nist.gov/) corpus with the aim of proactively anticipating threats. The use case provides the context for development, use and evaluation of the automated tools and processes based on them. The latter are incrementally and iteratively evaluated and improved as they are developed and used. Local evaluation and improvement come before global evaluation. We believe that performing global evaluation of text processing methodologies and processes currently exemplified by the Text REtrieval Conference (TREC), Document Understanding Conference (DUC) and Text Analysis Conference (TAC) are worth pursuing, and we have done so using TREC. However, we will mostly be focusing on local development, evaluation and improvement in this report. We will articulate various aspects of these approaches by describing and showing 1) our multi-word term based process for topic modeling that can be supported by semantic schema, taxonomic structures and knowledge graphs built out of the same multi-word terms; 2) the heuristic methods used to evaluate the performance of the multi-word term-based topic modeling that includes suggestions for measuring how well a topic is represented in the documents that are indexed to it; 3) how these local heuristic methods might be transformed into a new full-blown rigorous evaluation standard like TREC, DUC and TAC, but with emphasis on their contribution to interpretation and understanding of very large corpora. With respect to 3), we will also briefly explore what parts of the heuristic process would need to be automated and an indication of the algorithms needed to do so.

The Knowledge Graph for Macroeconomic Analysis with Alternative Big Data

DI-Mondrian: Distributed improved Mondrian for satisfaction of the L-diversity privacy model using Apache Spark

Low Resource Languages are the languages that few people speak and the low-density languages in the world. Usually, more people speak languages such as English, Chinese, Hindi, French, Spanish, etc. Researchers are more comfortable with the above languages. because these languages have annotated data or dataset available online for applying technologies like Voice assistance, Text summarization, Google translation, Auto correction, etc. Data availability provides good training and testing of the model. Most of the NLP applications are developed with high-resource languages only. Despite the advancement of the digital era, few languages have low density. One of the challenges of implementing NLP applications for low-resource languages are not having data, even if parsing tools are not available. This paper addressed a survey on improving NLP in low-resource languages, applications, and challenges.

The global impact of COVID-19 has been significant and several vaccines have been developed to combat this virus. However, these vaccines have varying levels of efficacy and effectiveness in preventing illness and providing immunity. As the world continues to grapple with the ongoing pandemic, the development and distribution of effective vaccines remains a top priority, making monitoring prevention strategies mandatory and necessary to mitigate the spread of the disease. These vaccines have raised a huge debate on social networks and in the media about their effectiveness and secondary effects. This has generated big data, requiring intelligent tools capable of analyzing these data in depth and extracting the underlying knowledge and feelings. There is a scarcity of works that analyze feelings and the prediction of these feelings based on their estimated polarities at the same time. In this work, first, we use big data and Natural Language Processing (NLP) tools to extract the entities expressed in tweets about AstraZeneca and Pfizer and estimate their polarities; second, we use a Long Short-Term Memory (LSTM) neural network to predict the polarities of these two vaccines in the future. To ensure parallel data treatment for large-scale processing via clustered systems, we use the Apache Spark Framework (ASF) which enables the treatment of massive amounts of data in a distributed way. Results showed that the Pfizer vaccine is more popular and trustworthy than AstraZeneca. Additionally, according to the predictions generated by Long Short-Term Memory (LSTM) model, it is likely that Pfizer will continue to maintain its strong market position in the foreseeable future. These predictive analytics, which uses advanced machine learning techniques, have proven to be accurate in forecasting trends and identifying patterns in data. As such, we have confidence in the LSTM's prediction of Pfizer's ongoing dominance in the industry.

The Smith-Waterman (SW) algorithm is universally used for a database search owing to its high sensitively. The widespread impact of the algorithm is reflected in over 8000 citations that the algorithm has received in the past decades. However, the algorithm is prohibitively high in terms of time and space complexity, and so poses significant computational challenges. Apache Spark is an increasingly popular fast big data analytics engine, which has been highly successful in implementing large-scale data-intensive applications on commercial hardware. This paper presents the first ever reported system that implements the SW algorithm on Apache Spark based distributed computing framework, with a couple of off-the-shelf workstations, which is named as SparkSW. The scalability and load-balancing efficiency of the system are investigated by realistic ultra-large database from the state-of-the-art UniRef100. The experimental results indicate that 1) SparkSW is load-balancing for parallel adaptive on workloads and scales extremely well with the increases of computing resource, 2) SparkSW provides a fast and universal option high sensitively biological sequence alignments. The success of SparkSW also reveals that Apache Spark framework provides an efficient solution to facilitate coping with ever increasing sizes of biological sequence databases, especially generated by second-generation sequencing technologies.

Linking event triggers with their respective arguments is an essential component for building an event extraction system. It is challenging to link event triggers with the corresponding arguments triggers when the sentence contains multiple events and arguments triggers. The task becomes even more challenging in a low-resource setup due to the unavailability of natural language processing resource and tools. In this paper, we study the event-argument linking task based on disaster event ontology in a low resource setup. We use BERT and non-BERT-based deep learning models in both monolingual and cross-lingual event-argument linking task.We also perform an ablation study of various features like position embeddings (PE), position indicator (PI), and segment ID (SI) to understand their contribution to performance improvement in non-BERT-based models. Using three different languages viz. Hindi, Bengali, and Marathi, we compare the results with multilingual BERT-based deep neural models in both monolingual and cross-lingual scenarios. We observe that the multilingual BERT-based model outperforms the best performing non-BERT-based model in cross-lingual settings. But in monolingual settings, the performance is similar in Hindi and Bengali datasets and slightly better in Marathi dataset. We choose the disaster domain due to its social implications. Our current experiments can be helpful in mining important information related to disaster events from news articles and building event knowledge graphs in low-resource languages.