Professional-grade software development is a key component of Elder Research's analytics consulting strategy. This area of expertise has matured into an award-winning capability in robust software application development, including commercial deployment of scientific and engineering solutions. Our team thrives on transforming innovative technology into useful software tools that solve meaningful problems. Our broad technology expertise facilitates finding optimal solutions and adapting to changing deployment environments. Our focus on software engineering best practices ensures that we efficiently deliver effective solutions.
Text Mining Tools
Text Mining is the science of leveraging textual data—from web pages to narrative fields within a database—for data mining. Text data is unstructured and challenging due to the richness and complexity of language; yet it holds enormous potential for forward-thinking firms due to its sheer volume and depth. These opportunities can usually be characterized as either Learning from Text or Process Automation.
Percept is a universal adapter to extract text from documents in a variety of formats including PDF, Word, and HTML and convert that text into a format suitable for predictive analytics by extracting semantic features. It was first introduced by Elder Research in February 2011 and has grown to include multiple text mining components created by the team. These components include:
- Extensible content extractor for reading multiple document formats and accessing filesystems, databases, and the web
- Basic text utilities for tokenization, stemming, and term substitution
- Statistical natural language processing engines including part of speech detection, phrase detection, and entity extraction
- Frequency-based, statistical multi-word phrase detector
- Efficient document similarity computation
- Outputs suitable for input into predictive analytics tools, open-source search engines, and tools proprietary to Elder Research such as DocGraph
Percept uses a pipeline-based architecture to process text documents in ways tailored to our clients’ needs.
DocGraph is a multi-dimensional similarity and visualization engine for exploring collections of text documents. Using a physics-based “spring” model, DocGraph intuitively displays documents close together to the degree that they share similar content. It supports keyword searching of documents and has options for filtering the search. It is available as a desktop application for Window, Linux and Mac platforms.
Glyphic is a library for the Java Virtual Machine enabling rule-based extraction of content from PDF documents. The name originates from the word “glyph”, a typographer's pictorial representation of a symbol. Many PDF extraction libraries focus only on the textual semantics of the documents. Glyphic employs that but also allows rule-based queries to use the structure and formatting of the document to identify possible content of interest. The library provides an expressive language for implementing content queries for PDF extraction, the data structures for efficient processing of those queries, and interfaces for accessing those queries from a larger system.
Scientific Software Engineering
The software development team’s mission is to partner with scientists, engineers, and academics to transform laboratory-grade software innovations into world-class, commercial-grade applications, because we believe that the impact of an innovation is proportional to its availability and usability.
Design Assessment of Reliability With INspection (DARWIN)
DARWIN is a tool for performing non-destructive analyses of crack formation and growth within turbine rotor disks (usually within aircraft turbines). Sponsored by the FAA and developed with Southwest Research Institute (SwRI) in collaboration with four major gas turbine engine manufacturers. It supports a variety of basic configuration combinations, allowing the user to specify:
- Probabalistic or deterministic analysis
- General or FAA-certification mode
- 1, 2, or 3 dimensional finite element models
Inherent or surface damage
- Automatic or manual crack growth zone specification
DARWIN predicts the fracture risk of components with undetected material anomalies or surface damage, and quantifies the influence of in-service inspection on risk reduction. DARWIN won an R&D 100 award in 2000, and is in wide use and further development to the present.
MicroFaVa is a high temperature (super-alloy) material property modeling tool for aerospace applications, designed to facilitate the efficient development and deployment of physics-based material property models. It was developed under NAVAIR SBIR Phase I & II programs in partnership with Southwest Research Institute, and is expanding under a NASA STTR program.
Numerical Evaluation of Stochastic Structures under Stress (NESSUS)
NESSUS® is a general-purpose tool for computing the probabilistic response of engineered systems. Developed with SwRI, it can be used to simulate uncertainties in loads, geometry, material behavior, and other user-defined random variables to predict the probabilistic response, reliability, and probabilistic sensitivity measures. NESSUS won an R&D 100 award in 2005, and is under continual use and improvement.