Berkant Barla Cambazoglu

Cold start link prediction

In the traditional link prediction problem, a snapshot of a social network is used as a starting point to predict, by means of graph-theoretic measures, the links that are likely to appear in the future. In this paper, we introduce cold start link prediction as the problem of predicting the structure of a social network when the network itself is totally missing while some other information regarding the nodes is available. We propose a two-phase method based on the bootstrap probabilistic graph. The first phase generates an implicit social network under the form of a probabilistic graph. The second phase applies probabilistic graph-based measures to produce the final prediction. We assess our method empirically over a large data collection obtained from Flickr, using interest groups as the initial information. The experiments confirm the effectiveness of our approach.

A large-scale sentiment analysis for Yahoo! answers

Sentiment extraction from online web documents has recently been an active research topic due to its potential use in commercial applications. By sentiment analysis, we refer to the problem of assigning a quantitative positive/negative mood to a short bit of text. Most studies in this area are limited to the identification of sentiments and do not investigate the interplay between sentiments and other factors. In this work, we use a sentiment extraction tool to investigate the influence of factors such as gender, age, education level, the topic at hand, or even the time of the day on sentiments in the context of a large online question answering site. We start our analysis by looking at direct correlations, e.g., we observe more positive sentiments on weekends, very neutral ones in the Science & Mathematics topic, a trend for younger people to express stronger sentiments, or people in military bases to ask the most neutral questions. We then extend this basic analysis by investigating how properties of the (asker, answerer) pair affect the sentiment present in the answer. Among other things, we observe a dependence on the pairing of some inferred attributes estimated by a users ZIP code. We also show that the best answers differ in their sentiments from other answers, e.g., in the Business & Finance topic, best answers tend to have a more neutral sentiment than other answers. Finally, we report results for the task of predicting the attitude that a question will provoke in answers. We believe that understanding factors influencing the mood of users is not only interesting from a sociological point of view, but also has applications in advertising, recommendation, and search.

Early exit optimizations for additive machine learned ranking systems

Some commercial web search engines rely on sophisticated machine learning systems for ranking web documents. Due to very large collection sizes and tight constraints on query response times, online efficiency of these learning systems forms a bottleneck. An important problem in such systems is to speedup the ranking process without sacrificing much from the quality of results. In this paper, we propose optimization strategies that allow short-circuiting score computations in additive learning systems. The strategies are evaluated over a state-of-the-art machine learning system and a large, real-life query log, obtained from Yahoo!. By the proposed strategies, we are able to speedup the score computations by more than four times with almost no loss in result quality.

Entity Recommendations in Web Search

While some web search users know exactly what they are looking for, others are willing to explore topics related to an initial interest. Often, the users initial interest can be uniquely linked to an entity in a knowledge base. In this case, it is natural to recommend the explicitly linked entities for further exploration. In real world knowledge bases, however, the number of linked entities may be very large and not all related entities may be equally relevant. Thus, there is a need for ranking related entities. In this paper, we describe Spark, a recommendation engine that links a users initial query to an entity within a knowledge base and provides a ranking of the related entities. Spark extracts several signals from a variety of data sources, including Yahoo! Web Search, Twitter, and Flickr, using a large cluster of computers running Hadoop. These signals are combined with a machine learned ranking model in order to produce a final recommendation of entities to user queries. This system is currently powering Yahoo! Web Search result pages.

A refreshing perspective of search engine caching

Commercial Web search engines have to process user queries over huge Web indexes under tight latency constraints. In practice, to achieve low latency, large result caches are employed and a portion of the query traffic is served using previously computed results. Moreover, search engines need to update their indexes frequently to incorporate changes to the Web. After every index update, however, the content of cache entries may become stale, thus decreasing the freshness of served results. In this work, we first argue that the real problem in todays caching for large-scale search engines is not eviction policies, but the ability to cope with changes to the index, i.e., cache freshness. We then introduce a novel algorithm that uses a time-to-live value to set cache entries to expire and selectively refreshes cached results by issuing refresh queries to back-end search clusters. The algorithm prioritizes the entries to refresh according to a heuristic that combines the frequency of access with the age of an entry in the cache. In addition, for setting the rate at which refresh queries are issued, we present a mechanism that takes into account idle cycles of back-end servers. Evaluation using a real workload shows that our algorithm can achieve hit rate improvements as well as reduction in average hit ages. An implementation of this algorithm is currently in production use at Yahoo!.

User engagement in online News: Under the scope of sentiment, interest, affect, and gaze

Online content providers, such as news portals and social media platforms, constantly seek new ways to attract large shares of online attention by keeping their users engaged. A common challenge is to identify which aspects of online interaction influence user engagement the most. In this article, through an analysis of a news article collection obtained from Yahoo News US, we demonstrate that news articles exhibit considerable variation in terms of the sentimentality and polarity of their content, depending on factors such as news provider and genre. Moreover, through a laboratory study, we observe the effect of sentimentality and polarity of news and comments on a set of subjective and objective measures of engagement. In particular, we show that attention, affect, and gaze differ across news of varying interestingness. As part of our study, we also explore methods that exploit the sentiments expressed in user comments to reorder the lists of comments displayed in news pages. Our results indicate that user engagement can be anticipated predicted if we account for the sentimentality and polarity of the content as well as other factors that drive attention and inspire human curiosity.

Machine learned job recommendation

We address the problem of recommending suitable jobs to people who are seeking a new job. We formulate this recommendation problem as a supervised machine learning problem. Our technique exploits all past job transitions as well as the data associated with employees and institutions to predict an employees next job transition. We train a machine learning model using a large number of job transitions extracted from the publicly available employee profiles in the Web. Experiments show that job transitions can be accurately predicted, significantly improving over a baseline that always predicts the most frequent institution in the data.

Quantifying performance and quality gains in distributed web search engines

Distributed search engines based on geographical partitioning of a central Web index emerge as a feasible solution to the immense growth of the Web, user bases, and query traffic. However, there is still lack of research in quantifying the performance and quality gains that can be achieved by such architectures. In this paper, we develop various cost models to evaluate the performance benefits of a geographically distributed search engine architecture based on partial index replication and query forwarding. Specifically, we focus on possible performance gains due to the distributed nature of query processing and Web crawling processes. We show that any response time gain achieved by distributed query processing can be utilized to improve search relevance as the use of complex but more accurate algorithms can now be enabled for document ranking. We also show that distributed Web crawling leads to better Web coverage and try to see if this improves the search quality. We verify the validity of our claims over large, real-life datasets via simulations.

Posting list intersection on multicore architectures

In current commercial Web search engines, queries are processed in the conjunctive mode, which requires the search engine to compute the intersection of a number of posting lists to determine the documents matching all query terms. In practice, the intersection operation takes a significant fraction of the query processing time, for some queries dominating the total query latency. Hence, efficient posting list intersection is critical for achieving short query latencies. In this work, we focus on improving the performance of posting list intersection by leveraging the compute capabilities of recent multicore systems. To this end, we consider various coarse-grained and fine-grained parallelization models for list intersection. Specifically, we present an algorithm that partitions the work associated with a given query into a number of small and independent tasks that are subsequently processed in parallel. Through a detailed empirical analysis of these alternative models, we demonstrate that exploiting parallelism at the finest-level of granularity is critical to achieve the best performance on multicore systems. On an eight-core system, the fine-grained parallelization method is able to achieve more than five times reduction in average query processing time while still exploiting the parallelism for high query throughput.

Query forwarding in geographically distributed search engines

Query forwarding is an important technique for preserving the result quality in distributed search engines where the index is geographically partitioned over multiple search sites. The key component in query forwarding is the thresholding algorithm by which the forwarding decisions are given. In this paper, we propose a linear-programming-based thresholding algorithm that significantly outperforms the current state-of-the-art in terms of achieved search efficiency values. Moreover, we evaluate a greedy heuristic for partial index replication and investigate the impact of result cache freshness on query forwarding performance. Finally, we present some optimizations that improve the performance further, under certain conditions. We evaluate the proposed techniques by simulations over a real-life setting, using a large query log and a document collection obtained from Yahoo!.