Giovanni Flammia

Multilingual spoken-language understanding in the MIT Voyager system

Abstract This paper describes our recent work in developing multilingual spoken language systems that support human-computer interactions. Our approach is based on the premise that a common semantic representation can be extracted from the input for all languages, at least within the context of restricted domains. In our design of such systems, language dependent information is separated from the system kernel as much as possible, and encoded in external data structures. The internal system manager, discourse and dialogue component, and database are all maintained in a language transparent form. Our description will focus on the development of the multilingual MIT Voyager spoken language system, which can engage in verbal dialogues with users about a geographical region within Cambridge, MA in the USA. The system can provide information about distances, travel times or directions between objects located within this area (e.g., restaurants, hotels, banks, libraries), as well as information such as the addresses, telephone numbers or location of the objects themselves. Voyager has been fully ported to Japanese and Italian, and we are in the process of porting to French and German as well. Evaluations for the English, Japanese and Italian systems are reported. Other related multilingual research activities are also briefly mentioned.

WebGALAXY: beyond point and click—a conversational interface to a browser

Abstract This paper presents WebGALAXY, a flexible multi-modal user interface system that allows wide access to selected information on the World Wide Web (WWW) by integrating spoken and typed natural language queries and hypertext navigation. WebALAXY extends our GALAXY spoken language system, a distributed client-server system for retrieving information from online sources through speech and natural language. WebGALAXY supports a spoken user interface via a standard telephone line as well as a graphical user interface via a standard Web browser using either Java/JavaScript or a cgi-bin/forms front end. Natural language understanding is performed by the system and information servers retrieve the requested information from various online resources including WWW servers, Gopher servers and CompuServe. Currently, queries about three domains are supported: weather, air travel, and points of interest around Boston.

Application service providers: challenges and opportunities

Application service provider is the new Internet buzzword. It describes a business model in which software is provided as an accessible service over distributed servers on the Internet. A growing number of applications (usually available as stand-alone programs on your computer), are also being made available as hosted subscription services. Hundreds of thousands of people use ASPs every day as an alternative to shrink-wrapped software. Established and emerging software companies are deploying such services. The ASP model introduces substantial changes in the way software is produced, used, and revised, and it forces us to rethink the whole software development process. The article provides some examples that demonstrate how widespread the model already is, and the challenges and opportunities this business model offers for computer scientists, software engineers, and designers of intelligent agent software.

Peer-to-peer is not for everyone

Software developers and specialized media are touting the peer-to-peer computing model (P2P) as the next big thing in Internet technology. I review and contrast it to the traditional Web-server model.

Speaker-independent consonant classification in continuous speech with distinctive features and neural networks

This paper provides experimental evidence to the assertion that the design of appropriate neural networks (NN) for speech recognition should be inspired by acoustic and phonetic knowledge, and not only by knowledge in pattern recognition. Rather than investigating the NN learning paradigm, the paper is focused on the influence of the input parameters, of the internal structure, and of the desired output representation on the classification performance of recurrent multilayer perceptrons. As an instructive example, the paper analyzes the problem of classifying ten stop and nasal consonants in continuous speech independently of the speaker. Experiments are reported for the TIMIT database, using 343 speakers in the training set and 77 different speakers in the test set. Comparative experiments show that good performance is obtained when many input acoustic parameters are used, including a time/frequency gradient operator related to transitions of the second formant, and when the desired outputs represent context-dependent articulatory features. Classification is performed by principal component analysis of the NN outputs. Refinement of the design parameters yield increasingly better performance on the test set, ranging from 45% errors for a perceptron without hidden nodes to 23.3% errors for the best NN.

XML and style sheets promise to make the web more accesible

The theme of the Sixth World Wide Web Conference, held in Santa Clara, April 6-11, was accessibility. Todays Hypertext Markup Language documents, with all the graphical bells and whistles added by Netscape and Microsoft, are biased toward a rich graphical user interface provided by the two most popular Web browsers running on desktop machines.

On the Internet, software should be milked, not brewed

The explosion of the Web raises many difficult legal and economic issues. One of the most controversial is this: how can we protect the intellectual property embedded in the computer software that gets distributed over the Internet? The Web hypertext platform, particularly hypertext transfer protocol (HTTP) and Java, should encourage the quick and ubiquitous delivery of novel, ground breaking ideas for interacting with computers. Innovative user interface software, whether in the form of Java byte codes or Web browsers, should be protected by law. But this is very difficult. Software products and ideas are too easily duplicated or cloned, and softwares authors are often compelled to expose the technical details of their products in order to disseminate them and create a market for their ideas and services, indirectly encouraging such cloning. The use of time stamping software to solve the Internet intellectual property problem is discussed.

From software to service: the transformation of shrink-wrapped software on the Internet

The paper considers how software is undergoing a fundamental change. A new more efficient kind of Internet-ready software is pre-empting traditional shrink-wrapped software. Consequently the software industry is quickly shifting from a product industry to a service industry. This transformation will make all aspects of software more efficient and is therefore inevitable. We have already seen hints of this future in both the corporate and research worlds. However, software designers will have to be fast and flexible to successfully compete in this new market. This new breed of software will bring new payment models as well. No longer will customers own software. Instead, they will rent software, based on their use.

Almost there: a new medium on the Web

In his famous article, Vannevar Bush (1945) envisioned memex (memory extension), a device in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility. Memex stores, links, and retrieves annotated text, photographs, speech, video, and film recordings. One of its important functions is that it allows annotation and retrieval, or selection of multimodal information on the basis of semantic content. According to Bush, to facilitate content retrieval, there is a new profession of trailblazers, those who find a delight in the task of establishing useful trails through the enormous mass of the common record. At the beginning of the article, he predicted that the interface to memex would use speech recognition and understanding. By the end, he envisioned an interface that could detect the electric waveforms from our brain to record thoughts and dreams and transform them into retrievable memex documents. In my opinion, the Web will eventually evolve into a truly dynamic multimodal medium without its current bias for static presentations. This medium will be different from and coexist with print, radio, cinema, television, cable, and video rentals. One of the features of memex that will materialize on the Web is the availability of audio and video media that can be linked and retrieved in much the same way as hypertext is.

A New Perspective on Peer-to-Peer

The author presents a new perspective on peer-to-peer computing inspired by the works of Digital Biology and Nexus. While their focus is clearly not peer-to-peer computing, their findings apply to it. They provide an introduction to fundamental concepts in networks that are applicable to Internet computing.