The paper’s problem of interest is the identification of an epidemic’s patient zero given a network of contacts and a set of infected individuals, under the assumptions that the infection states of only a few individuals are initially observed and the epidemic has evolved. To tackle this issue, they formulate the problem as an active querying problem and propose a number of active querying strategies inspired by active learning. Their results suggest that in the limited information scenario it is possible to achieve source inference performance comparable to when the infection states of all individuals are observed.

Current integration approaches are limited to two spaces and presume that both of them were embedded with the same method — both assumptions are unlikely to hold in the context of a Web of Embeddings. In this paper, we present FedCoder— an approach that integrates multiple embedding spaces via a latent space. We assert that linked entities have a similar representation in the latent space so that entities become comparable across embedding spaces. FedCoder employs an autoencoder to learn this latent space from linked as well as non-linked entities.

Our experiments show that FedCoder substantially outperforms state-of-the-art approaches when faced with different embedding models, that it scales better than previous methods in the number of embedding spaces, and that it improves with more graphs being integrated whilst performing comparably with current approaches that assumed joint learning of the embeddings and were, usually, limited to two sources. Our results demonstrate that FedCoder is well adapted to integrate the distributed, diverse, and large ecosystem of embeddings spaces into an interoperable Web of Embeddings.

You can read the full article here.

During May of this year, The Web Conference 2019 took place in San Francisco. Bibek Paudel, recent graduate of DDIS, was there to present a paper and poster titled ‘Iterative Learning Embeddings and Rules for Knowledge Graphs’. This paper was created in collaboration with —among others— visiting scholar Wen Zhang. We congratulate them on their successful work!

The paper presents an approach based on bayesian optimization that automatically tunes the myriad of parameters that a distributed stream processing system exposes. The paper will soon be available on our publication page.

A Clearly relaxed Cosmin after the defense…

TripleRush is a highly parallelized and distributed in-memory graph database (or triple store) that performs better than its competitors. It also has the special feature of being able to run sampling queries in the form of Random Walks with Restarts.

You can find all information about TripleRush in the paper, which is available at http://www.merlin.uzh.ch/publication/show/11663

Philip Stutz and Bibek Paudel presenting at WWW 2015

Newly minted PhD Philip Stutz wearing his hat!

Congratulations!

ps. Philip’s image was added to the DDIS portrait gallery.

First, the foundational paper of the Signal/Collect system was accepted by the Semantic Web Journal. The paper, provides a detailed description of Signal/Collect, introduces a number of algorithms, and provides a comparative scalability evaluation. You can find the latest draft here.

Second, a paper on TripleRush — a distributed, highly-parallel, in-memory Triplestore based on Signal/Collect  — was accepted to WWW 2015. A workshop version of the paper was presented at an ISWC 2012 workshop, which you can find here. We will post the final version in merlin as soon as we finished putting the final touches on the camera ready copy.

Congratulations to the authors Philip Stutz, Daniel Strebel, Bibek Paudel, and Ela Verman!