Coherence and Justification Clusters: Epistemic Metrics for Propositional Knowledge take from Interrogative Domains for Epistemic Agents

Abstract

Interrogative domains are domains where the primary activity consists of one or more people asking questions and one or more people giving answers. A civil or criminal trial is one example of an interrogative domain. Attorney's present questions to defendants, plaintiffs or witnesses, and the witnesses, plaintiffs, or defendants are charged with providing answers to the questions given. Other examples of an interrogative domain are congressional hearings, interviews and surveys. The question and answer sets taken from an interrogative domain are potential sources of knowledge. This potential is a result of the fact that given each question & answer pair entails at least one proposition. That proposition asserts something about the world that is either true or false. The agent(s), those individuals who are charged with making a determination or taking an action based on entailed propositions of the question and answer sets are candidate knowers in the interrogative domain. In other words based on the entailed propositions from the question and answer sets of the interrogative domains, what can the agent be said to know?

At Ctest Laboratories, we take the digital transcripts of interrogative domains, we extract the entailed propositions and then use mining algorithms and deductive/abductive inference analysis against those propositions to see if we can identify coherence and justification among the entailed propositions. That is, does any given proposition support any other given proposition in the transcript? How are the propositions in the transcript related or not related? Do any propositions challenge, impeach, or discredit any other propositions in the transcript? How can we characterize, organize and group these propositions and their relationships? We employ visualization techniques during the analysis that result in the propositions being placed into groups of various types and sizes of clusters we call coherence and justification clusters. We use the justification clusters as part of an epistemic metric that characterizes the integrity, quality and pedigree of any social knowledge that might be inferred from the entailed propositions that were taken from the original question & answer sets. If the justification clusters suggest that a set of propositions in the transcript pass the appropriate threshold of coherence then our epistemic agent is said to be committed to or (believe) that set of propositions. Further, if that same set of propositions pass an appropriate threshold of justification, we say that our epistemic agent is justified in its commitment to that set of propositions. With those two conditions in hand and some simplifying assumptions about the original entailed propositions, we talk about an epistemic agent having propositional knowledge taken from the interrogative domain, and we use justification clusters to characterize the validity of that knowledge. Our current research in computational epistemology is motivating us to believe that the digital transcripts of various interrogative domains are fertile sources of social knowledge. In this paper, we explore the use of coherence and justification clusters as a part of an epistemic metric that can validate whether or not, or to what degree social knowledge can be extracted from digital transcripts of interrogative domains for use by epistemic agents.

Discovering Coherence and Justification Clusters in Digital Transcripts Using Epistemic Analysis

Abstract

We are investigating the potential use of trial transcripts as sources of social knowledge for epistemic agents. But we are immediately faced with the reality that not all transcripts are equal. The quality of the transcripts will be partially related to the knowledge, consistency, and integrity of the individuals that testify during the course of the trial, and related to the nature and sophistication of the questions. Before we can determine whether a transcript will be useful as a knowledge source for an epistemic agent, we have to identify the consistency and quality of the knowledge present in the transcript. Coherence clusters demarcate the network of positively and negatively related propositions in the transcript. The justification clusters define the subcluster of propositions that support or justify other propositions in a coherence cluster. These clusters can be used to determine the nature of the consistency of the knowledge potentially present in the transcript. In this paper, we show how these clusters are identified using epistemic analysis. Our goals is to use these clusters as the basis for an epistemic metric used to determines the quality propositional knowledge present in a transcript.

Extracting Micro Ontologies from Interrogative Domains For Epistemic Agents

Abstract

A relevant and functional ontology continues to be one of the bottlenecks to the process of building epistemic agent-oriented systems. While the construction of electronic ontologies is the focus of many ongoing efforts, ontology building in a timely manner remains an obstacle. Our current focus is directed toward the notion of automated identification of ontological artifacts from interrogative domains in real-time. The artifacts that we are interested in form the basis for a micro-ontology of the interrogative domain under consideration.

Download Paper

Building A Priori Knowledge Spaces from Interrogative Domains

Abstract

A Priori knowledge is — the knowledge an agent has gained prior to experiences and learning. A priori knowledge acquisition along with relevant functional ontology building remain obstacles in the process of building knowledge-based agent-oriented systems. In this paper, we describe epistemic analysis techniques that we are exploring at Ctest Laboratories that are used to automatically discover ontological artifacts within the transcripts of interrogative domains. These ontological artifacts are then used as the fundamental basis of a priori knowledge spaces for epistemic agent-oriented systems. We are using ROGUE (Real-time Ontology Generation Using Epistemic Agents) to perform the transcript and text mining. ROGUE is a multi-agent system under development at Ctest Labs.

Download Paper

Transcript Mining Using Epistemic Agents and Interrogative Entailment

Abstract

Our current work is directed toward the mining and analysis of interrogative transcripts stored in digital form. In particular, we are interested in excavating the propositions and assertions that are implicit or entailed within the discourse of trial transcripts, law enforcement interrogations, congressional and other types of legal hearings. We are investigating the use of epistemic agents to mine and then perform epistemological analysis that can be used as the basis for understanding the consistency, validity, and soundness of the transcript as a whole. We use interrogative entailment as a mining process to qualify the credential of the transcript. In this paper, we describe the structure of our epistemic agents and the transcript mining process used to excavate statements that are entailed and inferred in the content of transcripts.

Download Paper

Epistemic Analysis of Interrogative Domain using Cuboids

Abstract

We are interested in analyzing the propositional knowledge extracted by an epistemic agent from interrogative domains. The interrogative domains that have our current focus are taken from transcripts of legal trials, congressional hearings, or law enforcement interrogations. These transcripts have be encoded in XML or HTML formats. The agent uses these transcripts as a primary knowledge source. The complexity, size, scope and potentially conflicting nature of transcripts from interrogative domains bring into question the quality of propositional knowledge that can be garnered by the agent. Epistemic Cuboids or Cubes are used as a knowledge analysis technique that helps determine the quality and quantity of the propositional knowledge extracted by an epistemic agent from an interrogative domain. In this paper we explore how ‘Epistemic Cubes’ can be used to evaluate the nature of the agent's propositional knowledge.

The Visualization of Epistemic Structures of Interrogative Domains

Abstract

At Ctest Laboratories we are investigating the concept of Epistemic Visualization of propositional knowledge in interrogative domains. Epistemic visualization is the process and result of developing visual models that capture the structure, content and justification of knowledge obtained by a rational software agent in a knowledge-based system. In this case the agen'ts knowledge is taken from electronic transcripts (e.g XML,HTML, plain-text formats, etc.) of interrogative domains such as trials, government hearings and interrogations. The transcripts are converted from their semi-structured text formats to an epistemic structure. The epistemic structure is a knowledge representation scheme used by the agent. In the instance of interrogative domains, the epistemic structure captures the propositional knowledge explicitly or implicitly contained in the transcripts. We present a visualization approach that graphically represents this epistemic structure using conventional visualization techniques.

Download Paper