Search NASA⌕ Search

SEARCH · Search NASA

Results for “semantic analysis”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
NASA NTRS2006

NASA Taxonomies for Searching Problem Reports and FMEAs

Many types of hazard and risk analyses are used during the life cycle of complex systems, including Failure Modes and Effects Analysis (FMEA), Hazard Analysis, Fault Tree and Event Tree Analysis, Probabilistic Risk Assessment, Reliability Analysis and analysis of Problem Reporting and Corrective Action (PRACA) databases. The success of these methods depends on the availability of input data and the analysts knowledge. Standard nomenclature can increase the reusability of hazard, risk and problem data. When nomenclature in the source texts is not standard, taxonomies with mapping words (sets of rough synonyms) can be combined with semantic search to identify items and tag them with metadata based on a rich standard nomenclature. Semantic search uses word meanings in the context of parsed phrases to find matches. The NASA taxonomies provide the word meanings. Spacecraft taxonomies and ontologies (generalization hierarchies with attributes and relationships, based on terms meanings) are being developed for types of subsystems, functions, entities, hazards and failures. The ontologies are broad and general, covering hardware, software and human systems. Semantic search of Space Station texts was used to validate and extend the taxonomies. The taxonomies have also been used to extract system connectivity (interaction) models and functions from requirements text. Now the Reconciler semantic search tool and the taxonomies are being applied to improve search in the Space Shuttle PRACA database, to discover recurring patterns of failure. Usual methods of string search and keyword search fall short because the entries are terse and have numerous shortcuts (irregular abbreviations, nonstandard acronyms, cryptic codes) and modifier words cannot be used in sentence context to refine the search. The limited and fixed FMEA categories associated with the entries do not make the fine distinctions needed in the search. The approach assigns PRACA report titles to problem classes in the taxonomy. Each ontology class includes mapping words - near-synonyms naming different manifestations of that problem class. The mapping words for Problems, Entities and Functions are converted to a canonical form plus any of a small set of modifier words (e.g. non-uniformity NOT + UNIFORM.) The report titles are parsed as sentences if possible, or treated as a flat sequence of word tokens if parsing fails. When canonical forms in the title match mapping words, the PRACA entry is associated with the corresponding Problem, Entity or Function in the ontology. The user can search for types of failures associated with types of equipment, clustering by type of problem (e.g., all bearings found with problems of being uneven: rough, irregular, gritty ). The results could also be used for tagging PRACA report entries with rich metadata. This approach could also be applied to searching and tagging failure modes, failure effects and mitigations in FMEAs. In the pilot work, parsing 52K+ truncated titles (the test cases that were available), has resulted in identification of both a type of equipment and type of problem in about 75% of the cases. The results are displayed in a manner analogous to Google search results. The effort has also led to the enrichment of the taxonomy, adding some new categories and many new mapping words. Further work would make enhancements that have been identified for improving the clustering and further reducing the false alarm rate. (In searching for recurring problems, good clustering is more important than reducing false alarms). Searching complete PRACA reports should lead to immediate improvement.

Malin, Jane T.↗
NASA NTRS2014

Aggregation Tool to Create Curated Data albums to Support Disaster Recovery and Response

Despite advances in science and technology of prediction and simulation of natural hazards, losses incurred due to natural disasters keep growing every year. Natural disasters cause more economic losses as compared to anthropogenic disasters. Economic losses due to natural hazards are estimated to be around $6-$10 billion dollars annually for the U.S. and this number keeps increasing every year. This increase has been attributed to population growth and migration to more hazard prone locations such as coasts. As this trend continues, in concert with shifts in weather patterns caused by climate change, it is anticipated that losses associated with natural disasters will keep growing substantially. One of challenges disaster response and recovery analysts face is to quickly find, access and utilize a vast variety of relevant geospatial data collected by different federal agencies such as DoD, NASA, NOAA, EPA, USGS etc. Some examples of these data sets include high spatio-temporal resolution multi/hyperspectral satellite imagery, model prediction outputs from weather models, latest radar scans, measurements from an array of sensor networks such as Integrated Ocean Observing System etc. More often analysts may be familiar with limited, but specific datasets and are often unaware of or unfamiliar with a large quantity of other useful resources. Finding airborne or satellite data useful to a natural disaster event often requires a time consuming search through web pages and data archives. Additional information related to damages, deaths, and injuries requires extensive online searches for news reports and official report summaries. An analyst must also sift through vast amounts of potentially useful digital information captured by the general public such as geo-tagged photos, videos and real time damage updates within twitter feeds. Collecting and aggregating these information fragments can provide useful information in assessing damage in real time and help direct recovery efforts. The search process for the analyst could be made much more efficient and productive if a tool could go beyond a typical search engine and provide not just links to web sites but actual links to specific data relevant to the natural disaster, parse unstructured reports for useful information nuggets, as well as gather other related reports, summaries, news stories, and images. This presentation will describe a semantic aggregation tool developed to address similar problem for Earth Science researchers. This tool provides automated curation, and creates "Data Albums" to support case studies. The generated "Data Albums" are compiled collections of information related to a specific science topic or event, containing links to relevant data files (granules) from different instruments; tools and services for visualization and analysis; information about the event contained in news reports, and images or videos to supplement research analysis. An ontology-based relevancy-ranking algorithm drives the curation of relevant data sets for a given event. This tool is now being used to generate a catalog of Hurricane Case Studies at Global Hydrology Resource Center (GHRC), one of NASA's Distribute Active Archive Centers. Another instance of the Data Albums tool is currently being created in collaboration with NASA/MSFC's SPoRT Center, which conducts research on unique NASA products and capabilities that can be transitioned to the operational community to solve forecast problems. This new instance focuses on severe weather to support SPoRT researchers in their model evaluation studies

Ramachandran, Rahul↗
NASA NTRS2013

Detecting and Characterizing Semantic Inconsistencies in Ported Code

Adding similar features and bug fixes often requires porting program patches from reference implementations and adapting them to target implementations. Porting errors may result from faulty adaptations or inconsistent updates. This paper investigates (I) the types of porting errors found in practice, and (2) how to detect and characterize potential porting errors. Analyzing version histories, we define five categories of porting errors, including incorrect control- and data-flow, code redundancy, inconsistent identifier renamings, etc. Leveraging this categorization, we design a static control- and data-dependence analysis technique, SPA, to detect and characterize porting inconsistencies. Our evaluation on code from four open-source projects shows thai SPA can dell-oct porting inconsistencies with 65% to 73% precision and 90% recall, and identify inconsistency types with 58% to 63% precision and 92% to 100% recall. In a comparison with two existing error detection tools, SPA improves precision by 14 to 17 percentage points

Ray, Baishakhi↗
NASA NTRS2009

Multi-user investigation organizer

A system that allows a team of geographically dispersed users to collaboratively analyze a mishap event. The system includes a reconfigurable ontology, including instances that are related to and characterize the mishap, a semantic network that receives, indexes and stores, for retrieval, viewing and editing, the instances and links between the instances, a network browser interface for retrieving and viewing screens that present the instances and links to other instances and that allow editing thereof, and a rule-based inference engine, including a collection of rules associated with establishment of links between the instances. A possible conclusion arising from analysis of the mishap event may be characterized as one or more of: not a credible conclusion; an unlikely conclusion; a credible conclusion; conclusion needs analysis; conclusion needs supporting data; conclusion proposed to be closed; and an un-reviewed conclusion.

Keller, Richard M.↗
DOE OSTI2024

Investigating Resilience of Loops in HPC Programs: A Semantic Approach with LLMs

Soft errors have become one of the major concerns for the error resilience of the HPC applications as those errors may cause HPC applications to generate serious outcomes such as silent data corruptions (SDCs). Protecting the applications from soft errors is an essential while challenging task. Among different approaches, obtaining a profound understanding of the resilience proneness of an application is very important to devise efficient error detection and recovery strategies. Given the scale of the HPC applications both in the code size and execution time, there are often cases that the error propagation analysis on such applications would produce a massive volume of unstructured data, which requires a significant amount of efforts, to process and to obtain indicating actions towards error protection. In this paper, we present a control-flow based visual analysis framework to help the users conduct error propagation analysis and identify the critical sections of a program that may have a higher likelihood of leading to erroneous outcomes when affected by the control flow related errors. We also design and implement the scalable visualization framework - ResilienceVis that efficiently and effectively visualizes the affected program states under errors and the propagation traces for an application in a user-friendly manner, and eventually, we combine the analysis and visualization to exhibit the error-proneness of the different sections of applications.

Jiang, Hailong↗
DOE OSTI2024

Automating the Analysis of Large Language Models Responses through Zero-Shot Question Answering

Recent advancements in Large Language Models (LLMs) have shown significant potential in various applications, yet their evaluation, particularly in zero-shot question answering scenarios, remains a challenging task. In this study, our objective was to explore precision metrics for Large Language Models (LLM) and design and implement a software pipeline to automatically evaluate LLMs' outputs under zero-shot question answering. Zero-shot question answering involves a model providing answers to questions about topics it hasn't seen during training. It leverages the principles of zero-shot learning by relying on semantic understanding and generalization from related knowledge. The data used was metadata from medical databases on congenital heart disease. We explored eleven LLM metrics and selected three for our evaluation: BLEU, BERTScore, and MoverScore. BLEU calculates a score based on the overlap of n-grams (contiguous sequences of n items, typically words) between the machine-generated translation and the reference translations. Higher BLEU scores indicate better correspondence between the machine-generated and human-generated translations. BERTScore is a metric used to evaluate the quality of machine-generated text by measuring the similarity of token embeddings produced by BERT (Bidirectional Encoder Representations from Transformers) between the generated text and reference text. MoverScore is a metric that quantifies the dissimilarity between the distributions of word embeddings from machine-generated text and reference text, emphasizing semantic similarity over exact token overlap. We also introduced HBKI, a composite metric summarizing these approaches. We tested five models —GPT-3, Llama-2, Gemini 1.5 Pro, Solar 10.7B, and Mixtral-8x7b. Our software pipeline, designed and implemented using Object-Oriented Programming principles, allows users to customize the selection and extraction of features for topics of interest in their own research. Our results show that MoverScore delivered the most precise evaluation of the LLM's outputs, while Mixtral-8x7b achieved the best overall performance in extracting metadata from the databases.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

AstraAI v1

AstraAI is an open-source, structure-aware AI coding agent designed for large scientific and DOE-HPC codebases such as AMReX-based applications. Unlike general-purpose coding assistants, AstraAI combines retrieval-augmented generation (RAG) with compiler-level Abstract Syntax Tree (AST) analysis to perform precise, scope-constrained code modifications. It identifies exact function spans, enforces locality of edits, and maintains cross-file invariants, enabling deterministic and build-safe transformations in complex C++/GPU environments. AstraAI is intended for developers working on large, evolving HPC frameworks where correctness, reproducibility, and structural integrity are critical. Typical use cases include modifying physics kernels, updating GPU device lambdas, and performing multi-file refactors without breaking compilation or runtime semantics. Compared to conventional LLM-based coding agents - even those with repository access - AstraAI provides structural guarantees rather than free-form text patches. It minimizes unintended diffs, prevents scope drift, preserves formatting and build stability, and reduces structural hallucinations. By integrating compiler tooling directly into the generation loop, AstraAI transforms AI-assisted coding from probabilistic text editing into deterministic, structure-preserving program transformation suitable for mission-critical scientific software.

Natarajan, Mahesh [Lawrence Berkeley National Labo↗
NASA NTRS2011

MOS 2.0: Modeling the Next Revolutionary Mission Operations System

Designed and implemented in the 1980's, the Advanced Multi-Mission Operations System (AMMOS) was a breakthrough for deep-space NASA missions, enabling significant reductions in the cost and risk of implementing ground systems. By designing a framework for use across multiple missions and adaptability to specific mission needs, AMMOS developers created a set of applications that have operated dozens of deep-space robotic missions over the past 30 years. We seek to leverage advances in technology and practice of architecting and systems engineering, using model-based approaches to update the AMMOS. We therefore revisit fundamental aspects of the AMMOS, resulting in a major update to the Mission Operations System (MOS): MOS 2.0. This update will ensure that the MOS can support an increasing range of mission types, (such as orbiters, landers, rovers, penetrators and balloons), and that the operations systems for deep-space robotic missions can reap the benefits of an iterative multi-mission framework.12 This paper reports on the first phase of this major update. Here we describe the methods and formal semantics used to address MOS 2.0 architecture and some early results. Early benefits of this approach include improved stakeholder input and buy-in, the ability to articulate and focus effort on key, system-wide principles, and efficiency gains obtained by use of well-architected design patterns and the use of models to improve the quality of documentation and decrease the effort required to produce and maintain it. We find that such methods facilitate reasoning, simulation, analysis on the system design in terms of design impacts, generation of products (e.g., project-review and software-delivery products), and use of formal process descriptions to enable goal-based operations. This initial phase yields a forward-looking and principled MOS 2.0 architectural vision, which considers both the mission-specific context and long-term system sustainability.

ground support systems↗
DOE OSTI2025

Can Large Language Models Understand Intermediate Representations?

Intermediate Representations (IRs) are essential in compiler design and program analysis, yet their comprehension by Large Language Models (LLMs) remains underexplored. This paper presents a pioneering empirical study to investigate the capabilities of LLMs, including GPT-4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama, in understanding IRs. We analyze their performance across four tasks: Control Flow Graph (CFG) reconstruction, decompilation, code summarization, and execution reasoning. Our results indicate that while LLMs demonstrate competence in parsing IR syntax and recognizing high-level structures, they struggle with control flow reasoning, execution semantics, and loop handling. Specifically, they often misinterpret branching instructions, omit critical IR operations, and rely on heuristic-based reasoning, leading to errors in CFG reconstruction, IR decompilation, and execution reasoning. The study underscores the necessity for IR-specific enhancements in LLMs, recommending fine-tuning on structured IR datasets and integration of explicit control flow models to augment their comprehension and handling of IR-related tasks.

Jiang, Hailong↗
NASA NTRSDate not supplied

A Provably Correct Floating-Point Implementation of Well Clear Avionics Concepts

The NASA DAIDALUS library provides formal definitions for Detect-and-Avoid avionics concepts such as when an aircraft is well-clear with respect to the surrounding air traffic, i.e., it does not operate in such proximity to create a collision hazard. While several properties are proven correct for DAIDALUS assuming ideal real number arithmetic, an actual implementation that uses floating-point numbers may behave unexpectedly because of round-off errors and run-time exceptions. This paper presents an experience report on the application of a formal methods toolchain to extract and verify floating-point C code from a real-valued specification of the well-clear module of DAIDALUS. This toolchain comprises the PVS theorem prover, the PRECiSA floating-point analyzer and code generator, and the Frama-C analysis suite. The generated code is automatically instrumented to detect when the control flow of the floating-point program may diverge from the ideal real number specification, and it is annotated with contracts that state the maximum accumulated round-off error. The absence of overflows is also formally verified for the generated code. In order to apply the toolchain to an industrial case study such as DAIDALUS, a formally verified pre-processing of the input specification is performed, which includes a program slicing and several semantic-preserving simplifications.

Program verification↗
NASA NTRS1992

Issues in knowledge representation to support maintainability: A case study in scientific data preparation

Scientific data preparation is the process of extracting usable scientific data from raw instrument data. This task involves noise detection (and subsequent noise classification and flagging or removal), extracting data from compressed forms, and construction of derivative or aggregate data (e.g. spectral densities or running averages). A software system called PIPE provides intelligent assistance to users developing scientific data preparation plans using a programming language called Master Plumber. PIPE provides this assistance capability by using a process description to create a dependency model of the scientific data preparation plan. This dependency model can then be used to verify syntactic and semantic constraints on processing steps to perform limited plan validation. PIPE also provides capabilities for using this model to assist in debugging faulty data preparation plans. In this case, the process model is used to focus the developer's attention upon those processing steps and data elements that were used in computing the faulty output values. Finally, the dependency model of a plan can be used to perform plan optimization and runtime estimation. These capabilities allow scientists to spend less time developing data preparation procedures and more time on scientific analysis tasks. Because the scientific data processing modules (called fittings) evolve to match scientists' needs, issues regarding maintainability are of prime importance in PIPE. This paper describes the PIPE system and describes how issues in maintainability affected the knowledge representation used in PIPE to capture knowledge about the behavior of fittings.

Chien, Steve↗
NASA NTRS1993

Intelligent assistance in scientific data preparation

Scientific data preparation is the process of extracting usable scientific data from raw instrument data. This task involves noise detection (and subsequent noise classification and flagging or removal), extracting data from compressed forms, and construction of derivative or aggregate data (e.g. spectral densities or running averages). A software system called PIPE provides intelligent assistance to users developing scientific data preparation plans using a programming language called Master Plumber. PIPE provides this assistance capability by using a process description to create a dependency model of the scientific data preparation plan. This dependency model can then be used to verify syntactic and semantic constraints on processing steps to perform limited plan validation. PIPE also provides capabilities for using this model to assist in debugging faulty data preparation plans. In this case, the process model is used to focus the developer's attention upon those processing steps and data elements that were used in computing the faulty output values. Finally, the dependency model of a plan can be used to perform plan optimization and run time estimation. These capabilities allow scientists to spend less time developing data preparation procedures and more time on scientific analysis tasks.

Chien, Steve↗
NASA NTRS1995

Formal methods in the design of Ada 1995

Formal, mathematical methods are most useful when applied early in the design and implementation of a software system--that, at least, is the familiar refrain. I will report on a modest effort to apply formal methods at the earliest possible stage, namely, in the design of the Ada 95 programming language itself. This talk is an 'experience report' that provides brief case studies illustrating the kinds of problems we worked on, how we approached them, and the extent (if any) to which the results proved useful. It also derives some lessons and suggestions for those undertaking future projects of this kind. Ada 95 is the first revision of the standard for the Ada programming language. The revision began in 1988, when the Ada Joint Programming Office first asked the Ada Board to recommend a plan for revising the Ada standard. The first step in the revision was to solicit criticisms of Ada 83. A set of requirements for the new language standard, based on those criticisms, was published in 1990. A small design team, the Mapping Revision Team (MRT), became exclusively responsible for revising the language standard to satisfy those requirements. The MRT, from Intermetrics, is led by S. Tucker Taft. The work of the MRT was regularly subject to independent review and criticism by a committee of distinguished Reviewers and by several advisory teams--for example, the two User/Implementor teams, each consisting of an industrial user (attempting to make significant use of the new language on a realistic application) and a compiler vendor (undertaking, experimentally, to modify its current implementation in order to provide the necessary new features). One novel decision established the Language Precision Team (LPT), which investigated language proposals from a mathematical point of view. The LPT applied formal mathematical analysis to help improve the design of Ada 95 (e.g., by clarifying the language proposals) and to help promote its acceptance (e.g., by identifying a verifiable subset that would meet the needs of safety-critical applications). The first LPT project, which ran from the fall of 1990 unti the end of 1992, produced studies of several language issues: optimization, sharing and storage, tasking and protected records, overload resolution, the floating point model, distribution, program erros, and object-oriented programming. The second LPT project, in 1994, formally modeled the dynamic semantics of a large part of the (almost) final language definition, looking especially for interactions between language features.

Guaspari, David↗
DOE OSTI2024

The Essence of Cryptol: A Denotational Cryptol Interpreter in Coq for Foundational Assurances for Quantum Resistant Cryptosystems

Systems of the utmost consequence need a means to establish authenticity of software and data. Cryptosystems implement authentication, but can be vulnerable to cryptographic and implementation attacks. With the threat of quantum cryptographic attacks, “post-quantum” cryptosystems (PQCs) must be henceforth used in these systems. However, the new cryptography needs new ways to, rigorously and machine-checkably, prove systems free of vulnerabilities. We propose a retargetable capability to rapidly instantiate proven correct postquantum cryptosystems through novel proof-carrying synthesis and proof-automation technique, extending those proven successful on existing systems. This capability is crucial to meeting the cryptographic requirements for future high-consequence systems. Since specifications for high consequence cryptography are presently captured in a domain specific language known as Cryptol. While this can enable convenient fully automated reasoning about Cryptol specificaitons and implementations via the Software Analysis Workbench (SAW), Cryptol has expressivity gaps, so that cryptosystems with probabilistic programming features like Falcon cannot be fully expressed in the language. Moreover, SAW’s automation fails for programs and specificaitons with inductive and recursive structure, as in the Sphincs+ PQC. Finally, Cryptol and SAW together represent some 200,000 lines of unverified Haskell, so that the any guarantees about high consequence cryptography are presently contingent on a large, unverified, yet trusted computing base. The first step of the larger project of agile, assured crpytography is therefore to provide a formal, mechanized semantics for Cryptol, so that the specifications expressed by cryptographers in Cryptol can be reasoned about and compiled into performant implementations with a foundational, machine checkable certificate of correctness. This report describes our work on this first step, culminating in the design of a certified denotational interpreter, in Coq, for core Cryptol.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

The CanBikeCO Full Pilot: Long-Term Results and Analysis From an E-Bike Program in Colorado, USA

Personal micromobility devices like bicycles, e-bikes, and scooters are low- or zero-energy alternatives to single-occupancy vehicles. However, a lack of data has led to a dearth of data-driven research on personally owned e-bike usage. We present longitudinal findings from the CanBikeCO program, focused on e-bike adoption and use across demographics, trip characteristics, and geographies in the state of Colorado. CanBikeCO recorded travel survey data from low-income individuals provided with personal e-bikes by the Colorado Energy Office in six communities across Colorado from July 2021 to December 2022. The data were collected using a custom instance of the National Renewable Energy Laboratory OpenPATH platform, which combines passive data collection with semantic information such as trip mode and purpose labels. To our knowledge, there are no prior travel survey data on personally owned e-bikes with this range and scope. Insights from this unique dataset include: (i) work trips were 17% more likely than average trips to be taken on an e-bike, (ii) e-bikes were most often reported to replace cars (34% of e-bike trips) and other personal micromobility devices (22%), and (iii) participants favored walking for trips less than 1 mile, e-bikes for trips of 1-3 miles, and e-bikes, cars, or shared rides for trips of 3-20 miles. The data used to generate these results have been made available in the Transportation Secure Data Center. We find e-bike use is appealing across age groups and may be related to characteristics of land use, urban form, occupation, income, and car ownership. We conclude for this population that the energy demand added by e-bike use (induced demand and replacing non-motorized modes) is outweighed by the reduction in energy demand from replacement of single-occupancy vehicle trips with e-bike trips. Our findings suggest considerable potential for energy savings from personal e-bike ownership.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2025

Automated model generation and parameter estimation of building energy models using an ontology-based framework

This study presents a methodology for automated model generation and parameter estimation of building energy models using semantic modeling and Bayesian estimation. Semantic modeling techniques are used to represent the system components and their interactions, facilitating the automatic generation of a simulation model from dynamic component models. The proposed approach is applied to a case study of a ventilation system where a simulation model is generated, calibrated, and assessed through different performance metrics. These metrics demonstrate the accuracy and reliability of both model point estimates and probabilistic prediction intervals across all model outputs. Overall, the proposed methodology offers a systematic and automated approach to model development and calibration in building energy systems, with potential applications in building performance analysis, monitoring, and optimization.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
NASA NTRS1994

Open architectures for formal reasoning and deductive technologies for software development

The objective of this project is to develop an open architecture for formal reasoning systems. One goal is to provide a framework with a clear semantic basis for specification and instantiation of generic components; construction of complex systems by interconnecting components; and for making incremental improvements and tailoring to specific applications. Another goal is to develop methods for specifying component interfaces and interactions to facilitate use of existing and newly built systems as 'off the shelf' components, thus helping bridge the gap between producers and consumers of reasoning systems. In this report we summarize results in several areas: our data base of reasoning systems; a theory of binding structures; a theory of components of open systems; a framework for specifying components of open reasoning system; and an analysis of the integration of rewriting and linear arithmetic modules in Boyer-Moore using the above framework.

Mccarthy, John↗
NASA NTRS2026

Machine Learning for Predicting Team Functioning in HERA Missions

Team functioning is integral to success in future long term space exploration missions. Proactively detecting declines in team functioning can mitigate conflict and ensure mission success. This project developed a speech-based artificial intelligence (AI) system that unobtrusively predicts degradation in team functioning, including performance and cohesion, in the Human Exploration Research Analog (HERA) Campaigns 4 and 5. The AI system conducted automated analysis of the prosodic (tone of voice) and linguistic (language content) components of speech, modeling interpersonal dynamics at both the turn-taking and day-wide levels. We investigated team functioning via observing structured interactions (i.e., multi-mission space exploration vehicle-extra vehicular activity [MMSEV-EVA], team interaction battery [TIB]) and unstructured interactions before the MMSEV-EVA task. We developed machine learning models to predict team functioning (objective task accuracy, self reported team efficacy and self reported team cohesion) by analyzing OpenSmile acoustic features, linguistic descriptors extracted via the linguistic inquiry and word count (LIWC) dictionary, and semantic embeddings. In the TIB, static models using logistic regression and random forests were not able to predict task accuracy, but predicted team efficacy and cohesion during both the decision making and relational tasks to a moderate level (60-70%). Majority voting on the individual turns to predict day long team efficacy further increased accuracies (70-80%). Finally, long short-term memory (LSTM) models showed the best performance across all variables (80-91%), including task performance. In the MMSEV-EVA, static models achieved an accuracy of 60% with majority voting, which increased to 80% through the incorporation of mission day as a variable, accounting for the learning effect. A key finding across both tasks was the "team-dependent" nature of these interactions; models achieved much higher accuracy when trained on prior days of the same team's data rather than attempting to generalize across entirely different teams, with even 1-2 days of prior data per team achieving 5-15% improvement over team-independent models. In addition, the incorporation of pre-task data from the same team also improves model performance, e.g., incorporating data from the decision-making task of the TIB, which preceded the relational task, improved the prediction of team efficacy and cohesion during the latter. We compared model performance when trained on machine-generated data compared to data that had been further corrected by human annotators. Overall, models trained on human-corrected data exhibited a modest improvement in performance, particularly when acoustic features were used. We found no significant correlation between word error rate (WER) and model accuracy (r(55) = -0.08, p = 0.51), but model’s accuracy was significantly higher for medium/high quality transcription (0.74 (SD = 0.48)) compared to the low-quality group (0.64 (SD = 0.36)) (t(63)=2.82, p = 0.006). Based on these, several design recommendation emerge, that could inform Standards at NASA. Models predicting team functioning should incorporate at least one to two days of historical interaction data, include brief pre-task discussions, and explicitly model temporal learning effects, especially for longer operational tasks. Minimum quality standards for automated speech-processing pipelines are needed, given the performance gains observed with manually corrected acoustic data. Finally, systems should leverage both acoustic features and language embeddings in complementary ways, with modality choices and fusion strategies tailored to mission context, task demands, and data quality requirements.

Shrivatsa Mishra↗