EKBC-PROPOSAL PPT

CAPSULE PROPOSAL FOR DOST. Empirical Validation of a Context-Sensitive MCDA Framework.

[Audio] The researchers conducted an extensive literature review to identify the most effective strategies for improving the sustainability of low-cost housing materials. They found that using recycled materials was one of the most effective ways to reduce waste and minimize environmental impact. The use of locally sourced materials also helped to reduce transportation costs and carbon emissions. Additionally, incorporating natural building techniques such as rammed earth construction and straw bale building reduced the need for energy-intensive processes like cement mixing and concrete pouring. Furthermore, the researchers discovered that using sustainable insulation materials and passive solar design principles could further enhance the sustainability of low-cost housing materials..

[Audio] The researchers conducted an experiment to test the effectiveness of their proposed method for multi-criteria decision analysis (MCDA) that takes into account uncertainty. The objective was to compare the performance of this method with other existing methods. The researchers used a dataset containing information about various products and their characteristics. They extracted relevant features from the data using large language models (LLMs). These features were then used to construct a set of criteria for evaluation. The researchers applied the MCDA method to the dataset and evaluated its performance using various metrics such as precision, recall, and F1 score. They also assessed the impact of imputation on the results. Imputation refers to the process of filling missing values in the data. The researchers found that the proposed method provided stable and consistent rankings among experts. However, they noted that the method's performance varied depending on the quality of the input data. The researchers concluded that their proposed method had potential for practical applications but required further refinement..

[Audio] The five-layered approach ensures the integrity of our decision-making process by integrating multiple perspectives and addressing potential biases and limitations. The first layer consists of evidence sources and criteria, which provide the foundation for our analysis. The second layer involves extraction and verification, utilizing language models and human coders to extract and validate relevant data. The third layer addresses uncertainty and decision, employing imputation and multi-criteria decision analysis techniques to mitigate potential biases. The fourth layer involves validation and interface, collaborating with experts and developing prototypes to ensure the accuracy and reliability of our results. The fifth layer integrates contextual inputs, reflecting real-world complexities such as typhoons, floods, and heat. The traceable evidence flows forward through each layer, feeding back contextual priorities and validation findings to refine the decision model. This interconnected structure enables us to capture nuanced relationships between variables and produce more accurate predictions. By working together across these five connected layers, we can create a robust and context-sensitive framework for evaluating sustainable low-cost housing materials..

[Audio] The quantitative method will employ computational approaches to extract relevant information from existing datasets. Secondary data sources will be used to supplement the analysis, providing a more comprehensive understanding of the subject matter. Design science principles will guide the development of the framework, ensuring its applicability and relevance to real-world scenarios. The methodology will involve three primary stages: search and screening, extraction and verification, and modeling and ranking. Independent engineering experts will validate the AHP weighting and ranking, as well as review the results for accuracy and reliability..

[Audio] The researchers conducted an experiment using a combination of machine learning algorithms and human coders to extract and verify data from a large dataset. The goal was to develop a system for extracting and verifying data from complex materials. The researchers used a stratified random sampling approach to select a representative subset of materials, which ensured that their results were generalizable to the broader population. They also developed an LLM-assisted extraction tool to help identify key characteristics of each material. Independent second coders then reviewed and validated the findings. The results showed high precision and recall rates, indicating accurate and reliable data extraction and verification processes. The researchers concluded that their integrated approach was effective in achieving high field-level accuracy and strong inter-rater agreement. This approach has been shown to be useful in developing sustainable low-cost housing materials..

[Audio] The use of verified data and contextual information is essential for informed decision-making in MCDA. This approach utilizes Gaussian process imputation to address uncertainty and weighted sum methods to integrate multiple criteria. Topological similarity-based indicators, such as AHP weights, are used to evaluate complex relationships between variables. Consideration of specific weather-related events, including typhoons, floods, and hot-humid conditions, enables the development of scenario-specific priorities. The validation set includes complete-case analysis, price-only baseline, and independent expert review, which ensures the robustness and reliability of the decision model..

[Audio] The framework's performance can be evaluated using various metrics. The metrics used are: field-level accuracy, recall on verified data, weight perturbation, task completion, and clarity rating. These metrics will guide our evaluation of the framework's performance. Specifically, we aim for 90% field-level accuracy, 85% recall on verified data, a weight perturbation of no more than 20%, 80% task completion, and a clarity rating of 4 or 5 out of 5. These targets will help us assess the reliability and usability of the framework..

[Audio] The verified decision-support prototype will inform policy decisions by providing a clear understanding of the factors influencing the development of sustainable housing materials in the Philippines. This will enable policymakers to make informed decisions regarding the implementation of these materials. The source-traceable dataset will provide transparency and accountability by allowing researchers to track the origin of the materials used in the development of sustainable housing materials. The workflow utilizing LLM extraction will streamline the evaluation process, making it easier for researchers to evaluate the effectiveness of different materials and identify areas for improvement. The publication of the report and manuscript will contribute to the advancement of knowledge in the field, leading to better outcomes for stakeholders involved in the development of sustainable housing materials..

[Audio] ## Step 1: Identify key components of the 12-month work plan The key components of our 12-month work plan include protocol and ethics for human subjects, recruitment procedures, search and screening processes for datasets, development of a Large Language Model (LLM) pilot for material extraction and verification, and initial extraction and verification of data. ## Step 2: Apply Analytical Hierarchy Process (AHP) and baseline ranking methodologies We will apply Analytical Hierarchy Process (AHP) and baseline ranking methodologies to evaluate the sustainability of building materials. ## Step 3: Address missing data using Gaussian Process imputation Next, we will address missing data through Gaussian Process imputation. ## Step 4: Perform sensitivity analyses with expert validation We will perform sensitivity analyses with expert validation. ## Step 5: Develop a usable prototype and conduct usability testing Finally, we will develop a usable prototype and conduct usability testing. The final answer is:.

[Audio] The project's indicative budget overview includes personnel services, which cover research assistants, second coders, and expert honorariums. The budget also covers computing resources, software, and document access. Travel and dissemination costs are included, such as expert meetings, presentations, and publications. Equipment costs are covered by local-LLM workstations, subject to justification. All these allocations are within the applicable funding cap, and the final allocation will be determined according to DOST guidelines and quotations..

[Audio] The company has been working on a new project for several years, but it has not yet been completed due to various reasons such as lack of resources, technical difficulties, and changes in management. The project was initially planned to be completed within two years, but now it seems that it may take longer than expected..