	Sign In Sign-Up

Jeffrey Goldman's Published Works

Complete Papers (as principle author)

A Digital Filter Model for Data Mining of Text Documents.
This is the abstract for my dissertation. It is being published by UMI publishing. My research work centers around discovering knowledge from free-form text documents.
Term Domain Distribution Analysis, a Data Mining Tool for Text Databases: A Case History in a Lung Cancer Text Database.
To appear in the 1998 1st International Discovery Science Conference This research is a short presentation of a text mining tool for knowledge discovery in document databases. It is a case study of thoracic lung cancer patients and the tumor laterality. This paper has a copyright from Springer.
Knowledge Discovery in an Earthquake Text Database: Correlation between Significant Earthquakes and the Time of Day
To appear in the 1997 SSDBM Conference. This research is a knowledge discovery case study of an earthquake text database. Using text parsing and counting methods, an interesting hypothesis emerged. The result was researched and verified with a statistical database. The conclusion demonstrates that earthquakes in California occur more often in the morning than any other time of the day.
Information Routing using a Corpus Distribution
To appear in the 1997 AAAI Conference. This extended abstract looks at the theoretical foundation of using a large collection of texts for Information Retrieval routing tasks. The approach uses statistics of individual words in comparison with a general collection.
On Using Logic Synthesis for Supervised Classification Learning (Co-author: Mark L. Axtell)
Appears in the 1995 Tools with AI Conference. This paper looks at experimental data on prediction performance for machine learning using logic synthesis tools. Specifically we compare Espresso, C4.5, and a generalized function decomposer. It was presented in November at the conference.
Pattern Theoretic Learning (Co-authors: Timothy Ross and David Gadd)
Appears in the 1995 AAAI Spring Symposium Series: Systematic Methods of Scientific Discovery. This paper explores using logic synthesis for scientific discovery in machine learning. Similar to the paper above, the focus is on demonstrating function decomposition's performance at learning over a broad band of concepts. The paper also shows the correlation of DFC and the number of samples needed to learn a particular function. The function classes include: Boolean, Numeric, Image, and String.
Pattern Theoretic Knowledge Discovery
Appears in the 1994 6th International IEEE Conference on Tools with Artificial Intelligence. This paper explores using logic synthesis as a tool for supervised classification learning and compares the relative performances of this method and C4.5. The test functions used are intended to represent database-like concepts.

Complete Papers (as secondary author)

Application of ESOP Minimization in Machine Learning and Knowledge Discovery
Appears in the 1995 International Workshop on Applications of Reed-Muller Expansion in Circuit Design. This paper explores Exorcism, an XOR sum of products logic minimization tool, as applied to Machine Learning problems.
Pattern Theoretic Feature Extraction and Constructive Induction
Appears in the 1994 ML-COLT Workshop on Constructive Induction. Its purpose is to describe function decomposition as a method for finding features in data.
On the Decomposition of Real-Valued Functions
This paper was presented at the (1994) 3rd International Workshop of Post-Binary ULSI Systems for an invited talk at the International Symposium on Multiple Valued Logic. The purpose of this paper is to show how function decomposition can be extended to decompose real-valued functions.

Technical Reports (as principle author)

The technical reports below are incomplete. The pages that are missing are the cover page, abstract page, and notice page. Also not included are the appendices. The reason they are not part of this reference is because those pages were not composed from LaTeX. Also, the appendices are typically very large (70+ pages). However, I do have paper copies available. Simply mail me to that affect and I will be happy so send you a complete copy of any of the technical reports.

Machine Learning: A Comparative Study of Pattern Theory and C4.5
A Comparison Technical Report of C4.5 and FLASH (additional graphs available upon request)
Expert Systems in Preprocessing: A Preliminary Study of Supervised Learning with Noise
Experiments with Noise Technical Report (additional graphs available upon request)
Pattern Theoretic Knowledge Discovery
A Technical Report version of the paper listed above (additional graphs available upon request)