This dissertation is about a particular style of research. The philosophy of this style is that in order to scientifically understand deep learning, it is fruitful to investigate what happens when neural networks are trained on simple, mathematically well-defined tasks. Even though the training data is simple, the training algorithm can end up…

In the United States, the internet is a vital resource for LGBTQ+ youth to meet their sexual and reproductive health information needs, especially those who cannot receive necessary information from family, healthcare providers, and classrooms. In this dissertation, I present three papers that connect content moderation policies and their impacts…

In response to the significant disruption posed by emergent AI technology, we propose a four part framework for teaching and learning practice and development. Rather than focus on the specific technologies of the moment, this framework provides actionable suggestions for individuals with varying views of AI and its positive and negative…

Offers characterization of sign-transmission which is more general than conventional signal-transmission theory. Concepts and terminology, formal description of individual communications process, reconciliation with classical signal-detection theory, applications of sign-detection formalism to information retrieval on MEDLINE database, and a…

Reports an approach to automatic thesaurus construction for Chinese documents. Presents an effective Chinese keyword extraction algorithm. Compared to previous studies, this method speeds up the thesaurus generation process drastically. It also achieves a similar percentage level of term relatedness. Includes three tables and four figures.…

Discusses search engines and provides methods for determining what resources are searched, the quality of the information, and the algorithms used that will improve the use of search engines on the World Wide Web, online public access catalogs, and electronic encyclopedias. Lists strategies for conducting searches and for learning about the latest…

Database Tomography (DT) is a system which includes algorithms for extracting multi-word phrase frequencies and performing phrase proximity analyses on any type of large textual database. DT was used to derive technical intelligence from a near-earth space (NES) database derived from the Science Citation Index and the Engineering Compendex.…

In information retrieval (IR), the aim of genetic algorithms (GA) is to help a system to find, in a huge documents collection, a good reply to a query expressed by the user. Analysis of phenomena seen during the implementation of a GA for IR has led to a new crossover operation, which is introduced and compared to other learning methods.…

As the use of online literature searching systems increases, the need for reducing the amount of nonrelevant material encountered by the users becomes more apparent. One method for achieving greater selectivity in document retrieval is by using relevance (or utility) estimates from past users to estimate document values for future users. However,…

This study found that a Kohonen self-organizing map (SOM) algorithm for browsing can categorize a large and eclectic Internet-information space into manageable sub-spaces that users can navigate to find relevant home pages. An automatically generated concept-space algorithm for searching was shown to enhance keyword-based Internet searching.…

Presents a probabilistic technique to retrieve passages from texts having a large size or heterogeneous semantic content. Results of experiments comparing the probabilistic technique to one based on a text segmentation algorithm revealed that the passage size affects passage retrieval performance; text organization and query generality may have an…

Proposes a new learning algorithm to improve the retrieval effectiveness of the search system used in the hypertext environment using an extended Boolean model with links to improve the ranking of retrieved items. Highlights include the basic retrieval scheme, the learning scheme, and a review of basic probabilistic retrieval models. (Contains 32…

Discusses methods of weighting of descriptors in document search patterns together with the concept of relative indexing and conditions which should be satisfied by descriptor weight. A procedure for modifiying document search patterns based on users' informational needs and opinions is outlined, and examples are given. Sixteen references are…

Presents an algorithmic approach to addressing the vocabulary problem in scientific information retrieval and information sharing, using the molecular biology domain as an example. A cognitive study and a follow-up document retrieval study were conducted using first a conjoined fly-worm thesaurus and then an actual worm database and the conjoined…

Presents methods of information query building in selective dissemination of information systems based on user's publications and examines possible compatibility between user's information queries and his publications. Methods of information query construction determined by logical operators AND, OR, NOT and set of weighted keywords are described.…