Russell Power

Transactional storage for geo-replicated systems

We describe the design and implementation of Walter, a key-value store that supports transactions and replicates data across distant sites. A key feature behind Walter is a new property called Parallel Snapshot Isolation (PSI). PSI allows Walter to replicate data asynchronously, while providing strong guarantees within each site. PSI precludes write-write conflicts, so that developers need not worry about conflict-resolution logic. To prevent write-write conflicts and implement PSI, Walter uses two new and simple techniques: preferred sites and counting sets. We use Walter to build a social networking application and port a Twitter-like application.

Transaction chains: achieving serializability with low latency in geo-distributed storage systems

Currently, users of geo-distributed storage systems face a hard choice between having serializable transactions with high latency, or limited or no transactions with low latency. We show that it is possible to obtain both serializable transactions and low latency, under two conditions. First, transactions are known ahead of time, permitting an a priori static analysis of conflicts. Second, transactions are structured as transaction chains consisting of a sequence of hops, each hop modifying data at one server. To demonstrate this idea, we built Lynx, a geo-distributed storage system that offers transaction chains, secondary indexes, materialized join views, and geo-replication. Lynx uses static analysis to determine if each hop can execute separately while preserving serializability---if so, a client needs wait only for the first hop to complete, which occurs quickly. To evaluate Lynx, we built three applications: an auction service, a Twitter-like microblogging site and a social networking site. These applications successfully use chains to achieve low latency operation and good throughput.

End-to-End Neural Ad-hoc Ranking with Kernel Pooling

This paper proposes K-NRM, a kernel based neural model for document ranking. Given a query and a set of documents, K-NRM uses a translation matrix that models word-level similarities via word embeddings, a new kernel-pooling technique that uses kernels to extract multi-level soft match features, and a learning-to-rank layer that combines those features into the final ranking score. The whole model is trained end-to-end. The ranking layer learns desired feature patterns from the pairwise ranking loss. The kernels transfer the feature patterns into soft-match targets at each similarity level and enforce them on the translation matrix. The word embeddings are tuned accordingly so that they can produce the desired soft matches. Experiments on a commercial search engines query log demonstrate the improvements of K-NRM over prior feature-based and neural-based states-of-the-art, and explain the source of K-NRMs advantage: Its kernel-guided embedding encodes a similarity metric tailored for matching query words to document words, and provides effective multi-level soft matches.

Explicit Semantic Ranking for Academic Search via Knowledge Graph Embedding

This paper introduces Explicit Semantic Ranking (ESR), a new ranking technique that leverages knowledge graph embedding. Analysis of the query log from our academic search engine, SemanticScholar.org, reveals that a major error source is its inability to understand the meaning of research concepts in queries. To addresses this challenge, ESR represents queries and documents in the entity space and ranks them based on their semantic connections from their knowledge graph embedding. Experiments demonstrate ESRs ability in improving Semantic Scholars online production system, especially on hard queries where word-based ranking fails.

Design and implementation of contextual information portals

This paper presents a system for enabling offline web use to satisfy the information needs of disconnected communities. We describe the design, implementation, evaluation, and pilot deployment of an automated mechanism to construct Contextual Information Portals (CIPs). CIPs are large searchable information repositories of web pages tailored to the information needs of a target population. We combine an efficient classifier with a focused crawler to gather the web pages for the portal for any given topic. Given a set of topics of interest, our system constructs a CIP containing the most relevant pages from the web across these topics. Using several secondary school course syllabi, we demonstrate the effectiveness of our system for constructing CIPs for use as an education resource. We evaluate our system across several metrics: classification accuracy, crawl scalability, crawl accuracy and harvest rate. We describe the utility and usability of our system based on a preliminary deployment study at an after-school program in India, and also outline our ongoing larger-scale pilot deployment at five schools in Kenya.

Oolong: asynchronous distributed applications made easy

We present Oolong, a distributed programming framework designed for sparse asynchronous applications such as distributed web crawling, shortest paths, and connected components. Oolong stores program state in distributed in-memory key-value tables on which user-defined triggers may be set. Triggers can be activated whenever a key-value pair is modified. The event-driven nature of triggers is particularly appropriate for asynchronous computation where workers can independently process part of the state towards convergence without any need for global synchronization. Using Oolong, we have implemented solutions for several large-scale asynchronous computation problems, achieving good performance and robust fault tolerance.