Search

In this chapter we present protocols for privately and securely matching two profiles (modelled as bit vectors), where a match means that the Hamming dis- tance between the profiles is sufficiently small. The protocol does not reveal any information other than the fact whether the two profiles match. We use the framework of secure computation based on threshold homomor- phic cryptosystems, as put forth by Cramer, Damg˚ ard, and Nielsen at Euro- crypt'01. More in particular, we use the extension introduced by Schoenmakers and Tuyls at Asiacrypt'04, which replaces the general multiplication gate by a restricted multiplication gate, called the conditional gate. The advantage of the conditional gate is that threshold homomorphic ElGamal suffices as the underly- ing cryptosystem, which allows for efficient discrete log based implementations, whereas a general multiplication gate requires the use of more involved RSA- like cryptosystems, such as Paillier's cryptosystem. The computational complexity of the protocol is dominated by the number of conditional gates needed. For profiles of length m, our protocol requires only O(m) conditional gates, where the hidden constant is small.

We study possibilities to automatically extract information from the Internet, by structuring and combining data from web pages. The web pages are found with the use of a search engine and the information is structured by using ontologies. The ontologies are populated with the use of statistical and linguistic techniques. We present the results of a case study that is aimed at finding the names of famous persons. The results indicate that, even if we only use the summaries that Google provides of web pages, the approach results in a high precision and recall for the specific application.

In this chapter we discuss multimodal interface technology. We present eexamples of multimodal interfaces and show problems and opportunities. Fusion of modalities is discussed and some roadmap discussions on research in multimodality are summarized. This chapter also discusses future developments where, rather tha communicating with a single computer, users communicate with their environment using multimodal interactions and where the environmental interface has perceptual competence that includes being able to interpret what is going on in the environment. We contribute roles to virtual humans in order to allow daily users of future computing environments to establish relationships with the environments, or more in particular, these virtual humans.