Ethel Ong

Planning Author and Character Goals for Story Generation

The design and content of the planning library of a story generation system dictates the content quality of the story it produces. This paper presents the story planner component of Picture Books, a system that generates stories for children aged 4 to 6 years based on a set of picture elements selected by the user. The planning library separates the design for the story patterns from the design of the semantic ontology that supplies the storys domain knowledge. An evaluation of the system shows that the coherency and completeness of the plot is attributed to the story pattern design structure while the appropriateness of the content is attributed to the semantic ontology.

Automatically Extracting Word Relationships as Templates for Pun Generation

Computational models can be built to capture the syntactic structures and semantic patterns of human punning riddles. This model is then used as rules by a computer to generate its own puns. This paper presents T-PEG, a system that utilizes phonetic and semantic linguistic resources to automatically extract word relationships in puns and store the knowledge in template form. Given a set of training examples, it is able to extract 69.2% usable templates, resulting in computer-generated puns that received an average score of 2.13 as compared to 2.70 for human-generated puns from user feedback.

Using common-sense knowledge in generating stories

A problem with most story generation systems is the lack of an adequately-sized body of knowledge to generate stories from. This paper presents an approach that focuses on providing a large amount of common-sense knowledge to automatic story generators while keeping extensive manual handcrafting of knowledge to a minimum. It does so by combining manually-created resources with freely-available common-sense knowledge in machine-readable format for the generation of stories.

Generating Punning Riddles from Examples

Text generation systems, such as pun generators, depend on manually created templates which require a lot of effort to build. This paper presents T-Peg, a system that utilizes semantic and phonetic knowledge sources to automatically capture the wordplay patterns of human-made jokes from training examples. The knowledge learned are stored as templates which, combined with a keyword input from the user, can then be used to generate punning riddles. Manual evaluation by a linguist on the completeness of the learned templates gave the system a score of 4.0 out of 5. User feedback gave the computer-generated puns an average score of 2.13 out of 5, as compared to their human-made counterparts which received an average score of 2.70, demonstrating that computers can be trained to be as humorous as humans.

Commonsense knowledge acquisition through children's stories

Humans interact with each other using their collection of commonsense knowledge about everyday concepts and their relationships. To establish a similar natural form of interaction with computers, they should be given the same collection of knowledge. Various research works have focused on building large-scale commonsense knowledge that computers can use. But capturing and representing commonsense knowledge into a machine-usable repository, whether manual or automated, are still far from completion. This research explores an approach to acquiring commonsense knowledge through the use of childrens stories. Relation extraction templates are also utilized to store the learned knowledge into an ontology, which can then be used by automatic story generators and other applications with children as the target users.

Planning children's stories using agent models

Automatic story generation systems must consider both consistency and coherency issues in the stories that they produce. This paper discusses the different agent models adapted to enhance the planning process of Picture Books 2 to produce stories that contain a logical flow of events. Along with the agent models, the ontology structure utilized as the source of storytelling knowledge must be efficiently designed to support the tasks of the agents. Preliminary results show that the interaction among the agents enabled the story planner to perform better than the earlier version of Picture Books 2 in generating logical stories.

Emotional Narration of Children’s Stories

Storytelling is both an entertaining and an educational tool for children. While Picture Books has shown the capability for a computer system to generate children’s stories, it lacks the storytelling capability needed in order to narrate the story to young beginning readers (age 4-6). Stories that are narrated with emotion can improve the experience of these young listeners. This paper presents the development of a synthesized voice narrator capable of expressing emotions while narrating the children’s stories generated by the Picture Books system. Test results showed that the children were able to recognize the generated sound clips that reflect the emotions happy and sad, but had difficulty recognizing the emotions reflecting anger, fear and surprise.

Building a Semantic Ontology for Virtual Peers in Narrative-Based Environments

Narrative-based environments utilize various forms of knowledge to provide an interactive space for the learner and the virtual agent to collaborate in accomplishing the learning goals. In this paper, we present the design of a semantic ontology that provides the necessary domain-based conceptual knowledge to allow a virtual peer to engage in storytelling as a form of exchange with the learner. We then show how the ontology was utilized to support the virtual peer in performing its tasks, which include generating interactive stories that teach about appropriate social behavior, and engaging in a text-based dialogue with the learner.

Building language resources for a Multi-Engine English-Filipino machine translation system

In this paper, we present the building of various language resources for a multi-engine bi-directional English-Filipino Machine Translation (MT) system. Since linguistics information on Philippine languages are available, but as of yet, the focus has been on theoretical linguistics and little is done on the computational aspects of these languages, attempts are reported here on the manual construction of these language resources such as the grammar, lexicon, morphological information, and the corpora which were literally built from almost non-existent digital forms. Due to the inherent difficulties of manual construction, we also discuss our experiments on various technologies for automatic extraction of these resources to handle the intricacies of the Filipino language, designed with the intention of using them for the MT system. To implement the different MT engines and to ensure the improvement of translation quality, other language tools (such as the morphological analyzer and generator, and the part of speech tagger) were developed.

Building a Commonsense Knowledge Base for a Collaborative Storytelling Agent.

Storytelling is a common activity that people engage in to share and exchange information about their everyday life events. Children in particular find storytelling entertaining as they learn about their world and even share their own stories. Virtual agents are gaining popularity as conversational agents who can engage with their human users in a dialogue to answer queries and to find the necessary support in the performance of some tasks in a particular application domain. For virtual agents to be able to share stories with their human users, specially the children, they need to be provided with a body of knowledge to use in their dialogue. In this paper, we describe our approach in building an ontology to provide a knowledge base of commonsense concepts and their relations to a virtual agent. This is then used by a virtual agent to process user input as a form of story text, and to generate appropriate responses in order to encourage the child to share his/her story.