Robert Wilensky

Talking to UNIX in English: an overview of UC

UC is a natural language help facility which advises users in using the UNIX operating system. Users can query UC about how to do things, command names and formats, online definitions of UNIX or general operating systems terminology, and debugging problems in using commands. UC is comprised of the following components: a language analyzer and generator, a context and memory model, an experimental common-sense planner, highly extensible knowledge bases on both the UNIX domain and the English language, a goal analysis component, and a system for acquisition of new knowledge through instruction in English. The language interface of UC is based on a “phrasal analysis” approach which integrates semantic, grammatical and other types of information. In addition, it includes capabilities for ellipsis resolution and reference disambiguation.

The berkeley UNIX consultant project

UC (UNIX Consultant) is an intelligent, natural language interface that allows naive users to learn about the UNIX2 operating system. UC was undertaken because the task was thought to be both a fertile domain for artificial intelligence (AI) research and a useful application of AI work in planning, reasoning, natural language processing, and knowledge representation.The current implementation of UC comprises the following components: a language analyzer, called ALANA, produces a representation of the content contained in an utterance; an inference component, called a concretion mechanism, that further refines this content; a goal analyzer, PAGAN, that hypothesizes the plans and goals under which the user is operating; an agent, called UCEgo, that decides on UCs goals and proposes plans for them; a domain planner, called KIP, that computes a plan to address the users request; an expression mechanism, UCExpress, that determines the content to be communicated to the user, and a language production mechanism, UCGen, that expresses UCs response in English.UC also contains a component, called KNOME, that builds a model of the users knowledge state with respect to UNIX. Another mechanism, UCTeacher, allows a user to add knowledge of both English vocabulary and facts about UNIX to UCs knowledge base. This is done by interacting with the user in natural language.All these aspects of UC make use of knowledge represented in a knowledge representation system called KODIAK. KODIAK is a relation-oriented system that is intended to have wide representational range and a clear semantics, while maintaining a cognitive appeal. All of UCs knowledge, ranging from its most general concepts to the content of a particular utterance, is represented in KODIAK.

Meta‐Planning: Representing and Using Knowledge About Planning in Problem Solving and Natural Language Understanding*

This paper is concerned with those elements of planning knowledge that are common to both understanding someone elses plan and creating a plan for ones own use. This planning knowledge can be divided into two bodies: Knowledge about the world, and knowledge about the planning process itself. Our interest here is primarily with the latter corpus. The central thesis is that much of the knowledge about the planning process itself can be formulated in terms of higher-level goals and plans called meta-goals and meta-plans. These entities can then be used by the same understanding and planning mechanisms that process ordinary goals and plans. However, the meta-planning knowledge now enables these mechanisms to handle much more complicated situations in a uniform manner. Systems based on meta-planning would have a number of advantages over existing problem solving and understanding systems. The same knowledge could be shared by both a planner and understander, and both would be able to handle complex situations elegantly. In addition, in planning, the use of metaplanning has several advantages over more traditional methods involving constraints or critics. Meta-planning allows the full power of a problem solver to be applied to situations that are generally amenable only to special purpose processing. In addition, meta-planning facilitates the representation of some situations that are difficult to express otherwise. We have begun to introduce metaplanning knowledge into two systems: PAM, a story understanding program, and PANDORA, a problem solving and planning system.

Robust intra-document locations

Abstract Several types of existing and next-generation hypertext functionality, including external hyperlinks, annotations, and transclusions, rely on references to locations within another resource. If the document domain cannot guarantee referential integrity, but rather is more like the World Wide Web, in which documents change regularly and without notification, potentially invalidating internal location references, it is crucial to build robustness into the intra-document location resolution mechanism, so that locations continue to function even as documents change chaotically. This paper aims to begin a process to evolve a standard for (normative) robust location descriptors and (non-normative) reattachment algorithms. We discuss criteria for evaluating the robustness of an intra-document location mechanism. Then we describe robust locations , an approach we believe meets these criteria. Robust locations include a standard minimal location descriptor and a recommended reattachment algorithm. We also suggest what can be done when the changes are so great that location resolution is problematic. Finally, we describe the implementation of robust locations within the Multivalent Document system.

An algorithm for automated rating of reviewers

The current system for scholarly information dissemination may be amen able to significant improvement. In particular, going from the current system of journal publication to one of self-distributed documents offers significant cost and timeliness advantages. A major concern with such alternatives is how to provide the value currently afforded by the peer review system. Here we propose a mechanism that could plausibly supply such value. In the peer review system, papers are judged meritorious if good reviewers give them good reviews. In its place, we propose a collaborative filtering algorithm which automatically rates reviewers, and incorporates the quality of the reviewer into the metric of merit for the paper. Such a system seems to provide all the benefits of the current peer review system, while at the same time being much more flexible. We have implemented a number of parameterized variations of this algorithm, and tested them on data available from a quite different application. Our initial experiments suggest that the algorithm is in fact ranking reviewers reasonably.

Toward active, extensible, networked documents: multivalent architecture and applications

Rich varieties of online digital documents are possible, documents which do not merely imitate the capabilities of other media. A true digital document provides an interface to potentially complex content. Since this content is infinitely varied and specialized, we must provide means to interact with it in arbitrarily specialized ways. Furthermore, since relevant content may be found in distinct documents, we must draw from multiple sources, yet provide a coherent presentation to the user Finally, it is essential to be able to conveniently author new content, dejine new means of manipulation, and seamlessly mesh both with existing materials. Wepresent a new general paradigm that regards documents with complex content as “multivalent documents”, comprising multiple “layers” of distinct but intimately related content. Small, dynamically-loaded program objects, or “behaviors”, activate the content and work in concert with each other and layers of content to support arbitrarily specialized document types. Behaviors bind together the disparate pieces of a multivalent document to present the user with a single untjied conceptual document. As implemented in Java in the context of the World Wide Web, multivalent documents in effect create a customizable virtual Web, drawing together diverse content and functionality into coherent documentbased inte~aces to content. Examples of the diverse jimctionality in multivalent documents include: “OCR select and paste”, where the user describes a geometric region on the scanned image of a printed page and the corresponding text characters are copied out; video subtitling, which aligns a video clip with the script and language translations so that, e.g., the playing video can be presented simultaneously in multiple languages, and the video can be searrhed with text-based techniques; geographic information system (GIS) visualizations that compose several types of data from multiple datasets; and distributed user annotations that augment and may transform the Content of other documents. In general, a document management infrastructure built around a multivalent perspective can provide an extensible, networked system that supports incremental addition of content, incremental addition of interaction with the user and with other components, reuse of content across behaviors, reuse of behaviors across types of documents, and eficient use of network bandwidth. Multivalent The work reported here was supported in part by National Science Foundation grant IRI-9411334 as part of the NSF/NASA/ARPA Digital

Toward work-centered digital information services

Work-centered digital information services are a set of library-like services meant to address work group needs. Workplace users especially need to access legacy documents and external collections. They also frequently want to retrieve information (rather than documents per se), and they require that digital information systems be integrated into established work practices. Realizing work-centered digital information systems requires a broad technical agenda. Three types of analysis-document image, natural language, and computer vision, are necessary to facilitate information extraction. Users also need new user interface paradigms and authoring tools to better access multimedia information, as well as improved protocols for client-program interaction with repositories (collections). Moreover, entirely new types of documents must be developed to exploit these capabilities. The system developed by the authors follows a client-server architecture, in which the servers are repositories implemented as databases supporting user-defined functions and user-defined access methods. The repositories also serve as indexing servers. The authors are creating a prototype set of information services called the California Environmental Digital Information System, which includes a diverse collection of environmental data.

The hash history approach for reconciling mutual inconsistency

We introduce the hash history mechanism for capturing dependencies among distributed replicas. Hash histories, consisting of a directed graph of version hashes, are independent of the number of active nodes but dependent on the rate and number of modifications. We present the basic hash history scheme and discuss mechanisms for trimming the history over time. We simulate the efficacy of hash histories on several large CVS traces. Our results highlight a useful property of the hash history: the ability to recognize when two different non-commutative operations produce the same output, thereby reducing false conflicts and increasing the rate of convergence. We call these events coincidental equalities and demonstrate that their recognition can greatly reduce the time to global convergence.

Digital library resources as a basis for collaborative work

The creation of large, networked, digital document resources has greatly facilitated information access and dissemination. We suggest that such resources can further enhance how we work with information, namely, that they can provide a substrate that supports collaborative work. We focus on one form of collaboration, annotation, by which we mean any of an open‐ended number of creative document manipulations that are useful to record and to share with others. Widespread digital document dissemination required technological enablers, such as web clients and servers. The resulting infrastructure is one in which information may be widely shared by individuals across administrative boundaries. To achieve the same ubiquitous availability for annotation requires providing support for spontaneous collaboration, that is, for collaboration across administrative boundaries without significant prior agreements. Annotation is not more commonplace, we suggest, because the technological needs of spontaneous collaboration are challenging. We have developed a document model, called multivalent documents, which provides a means to address these challenges. In the multivalent document model, a document comprises distributed data and program resources, called layers and behaviors, respectively. Because most document functionality is implemented by behaviors, the model is highly extensible, and can accommodate both new document formats and novel forms of functionality. Among other applications, it is possible to use the model to effect a wide class of annotation types, across different document formats, without any administrative provisions. An implementation of the model has allowed us to develop behaviors that currently support some quite different but common digital document types, and a number of quite different annotation capabilities—some familiar, and some novel. A related implementation provides some analogous capabilities for geographic data. Such capabilities could have a beneficial impact on the “scholarly information life cycle,” i.e., the process by which researchers and scholars create, disseminate, and use knowledge.

PHRAN - A Knowledge-Based Natural Language Understander

We have developed an approach to natural language processing in which the natural language processor is viewed as a knowledge-based system whose knowledge is about the meanings of the utterances of its language. The approach is oriented around the phrase rather than the word as the basic unit. We believe that this paradigm for language processing not only extends the capabilities of other natural language systems, but handles those tasks that previous systems could perform in a more systematic and extensible manner.We have constructed a natural language analysis program called PHRAN (PHRasal ANalyzer) based in this approach. This model has a number of advantages over existing systems, including the ability to understand a wider variety of language utterances, increased processing speed in some cases, a clear separation of control structure from data structure, a knowledge base that could be shared by a language production mechanism, greater ease of extensibility, and the ability to store some useful forms of knowledge that cannot readily be added to other systems.