Perilog:
Contextual Search and Retrieval Software Tools
Objective
The National Aeronautics and Space Administration seeks companies to commercialize Perilog. Perilog is a suite of data mining tools that retrieves and organizes contextually relevant data from any sequence of terms (text, musical notes, genetic data, etc.). It is an integrated set of methods that can be used to intelligently mine information from databases.
Product Profile
Perilog designed originally to support the FAA’s Aviation Safety Reporting System (ASRS). The ASRS test bed demonstrated Perilog’s power on a topical database of thousands of documents. The algorithm was powerful enough to produce the first quantitative evidence of situational relationships between reported commercial aviation incidents and a specific type of aviation accident (McGreevy & Statler, 1998).
Perilog unearths data that is contextually relevant to the subject being investigated. The software measures the degree of contextual association for large numbers of term pairs in text (or any other sequence) to produce models that capture the structure of the text. Perilog statistically compares these models to measure their degree of similarity to a query model, develops a ranking, and presents the search results to the user. Furthermore, the user has access to powerful query tools that, for example, generate search options automatically.
Technical Basics
Perilog relies on four methods for data mining:
- Keyword-in-Context Search retrieves narratives that contain one or more user-specified keywords in typical or selected contexts, and ranks the narratives on their relevance to the keywords in context.
- Flexible, Model-Based Phrase Search retrieves narratives that contain one or more user-specified phrases, and ranks the narratives on their relevance to the phrases.
- Model-Based Phrase Generation produces a list of phrases from documents that contain a user-specified word or phrase.
- Narrative-Based Phrase Discovery finds phrases that are related to topics of interest by generating a list of narratives similar in meaning to the keyword or phrase query.
Perilog methods derive from seven inventive concepts:
- Modeling contextual associations (directional and non-directional) in sequences
- Deriving a query model from key terms and a database of contextual models
- Key term-in-context search
- Model-based phrase search
- Generating phrases from contextual models of sequences
- Extracting phrases from sequences, selecting relevant phrases
- Discovering associated phrases by iterative extraction and selection
Perilog offers the following features:
Analysis:
Measures contextual associations within sequences
Relevance-ranking:
Ranks collections of contexts on similarity to other collections of
contexts.
Search:
Provides more effective key term and phrase search.
Phrase mining:
Helps users to know what phrases occur in a database, and discovers
phrases that are contextually related to key terms or phrases.
Modeling:
Represents contexts within sequences as pairwise inter-term contextual
associations having quantified degrees of association.
- Generates network representations of collections of contexts.
- Models represent contexts of both real-world and symbolic terms.
- The contextual scope of models is easy to manipulate.
Potential Commercial Uses
Perilog has applications in a multitude of fields. Any knowledge management application can benefit from the technology. Perilog’s manipulation of patterned or sequential symbols, data, items, objects, events, causes, time spans, actions, attributes, entities, relations and representations allows powerful search and mining of any type of information repository.
BIOTECH: Perilog may be employed in biotechnology to unearth contextually relevant information, data, representations, sequences, and patterns. It may also be used to mine and retrieve genetic and protein sequences, representations, and analogs.
MULTIMEDIA: Perilog may also be used to manipulate multimedia repositories for “smart” retrieval of sound, music, voice, audio data, audio encoding, and vocal encoding.
CORPORATE NETWORKS: In a corporate text-based repository environment, Perilog algorithms may be applied to any form of text; for example, narratives, reports, literature, patents, punctuation, messages, electronic mail, internet text, and web site information and URLs. Additionally, it processes linguistic patterns and grammatical tags. Perilog is able to streamline any large topical database, manage corporate-wide knowledge repository (extendable to nodes outside the corporation via network), and push information based on specific research, marketing, or organizational projects to querying users.
Benefits
Perilog has tangible benefits such as context-sensitivity and “smart” context ranking. It can distill topically and situationally relevant phrases and deliver relevant results to a user who is not a subject expert. Furthermore, Perilog does not require manipulation of the subject database or expert intervention to achieve results.
| Other Search Technology | Perilog Innovations |
|---|---|
| Based on isolated term occurrence or linguistic patterns. | Based on structure of domains, situations, concerns and narratives |
| Term Indexing | Contextual indexing |
| Treats phrases as terms | Models intra-phrase relations |
| list frequent noun phrases | Represents all phrases |
| Handmade tables | Self-generated Tables |
Technology Commercialization Status
Perilog consists of modules embodied in four patents (pending). NASA is seeking non-exclusive license applicants.
Contact Information
If your company is interested in this technology or would like licensing instructions, please contact:
Marty Zeller
NASA Far West Technology Transfer Center
3716 S Hope St. #200
Los Angeles, CA 9007
Phone: (213) 743-2353
Fax: (213) 746-9043
E-mail: zeller@usc.edu
More information can be found at the Perilog Informational Event website at:
http://ettc.usc.edu/ames/perilog
