International Business Machines Corporation

Microsoft Technology Licensing, LLC

Yuanyuan Tian

Shirish Tatikonda

Berthold Reinwald

Andrey Balmin

Vijayshankar Raman

Ronald Jason Barber

Mir Hamid Pirahesh

Richard Sefton Sidle

John Sismanis

Adam J Storm

Shivakumar Vaithyanathan

Ronen Grosman

Douglas Ronald Burdick

Pinar Tozun

Guy Maring Lohman

Peter Haas

Christian Marcelo Garcia-Arellano

Prithviraj Sen

Rene Mueller

Matthias Boehm

Jyoti Leeka

Wenlei Xie

Bin Bi

Wen Sun

Fatma Ozcan

Edwin Peter Dawson Pednault

Rajasekar Krishnamurthy

Alekh Jindal

Carlo Aldo Curino

Vikas Sindhwani

John A McPherson, Jr

Tarique Ashraf Siddiqui

Chen Luo

Hamid Pirahesh

Amol N Ghoting

Sui Jun Tong

Brandon Barry Haynes

Matthew J Spilchen

Jesus Camacho Rodriguez

Anna Pavlenko

Rana Alotaibi

Hanxian Huang

Rana Bijad M Alotaibi

Tao Zou

Severin A Corsten

Examples detect equivalent subexpressions within a computational workload. Examples include converting a query plan tree associated with a first subexpression into a matrix. The first subexpression is a portion of a database query from the computational workload. Each node in the query plan tree is represented as a row of the matrix. The matrix is converted into a first vector. The first subexpression is determined to be equivalent to a second subexpression by comparing the first vector to a second vector associated with the second subexpression. The comparison includes computing a distance between the first and second vectors that is lower than a distance threshold. The computational workload is modified, based on the determining, to perform the first subexpression and exclude performance of the second subexpression as duplicative.

Machine learning accelerated semantic equivalence detection

Systems and methods for characterizing and forecasting evolving query workloads. The method includes receiving a query, the received query including a parameter value and an arrival time; identifying the query as a recurrent query; extracting a query template from the received query by parsing the received query; based at least on the identifying, generating a feature vector for the received query, the feature vector generated based on the extracted template and the parameter value; and forecasting a future query based on the generated feature vector by applying a neural network.

Characterizing and forecasting evolving query workloads

Embodiments relate to aggregation operations. More specifically, embodiments relate to enforcing an aggregation constraint across a distributed network of compute nodes responding asynchronously. Each compute node individually processes the local partial aggregate value in parallel and asynchronously communicates with a conflict resolution node (CRN).

Scalable enforcement of aggregation constraints within transactions

One embodiment provides for a method including performing, by a processing thread, a process that analyzes transactional operations by maintaining the transactional operations in transaction local side logs, and waiting until a successful transaction commit to append the transaction local side logs to a log stream. The processing thread processes the transactional operations on a key used to determine whether existing data is found for the key. The transactional operations are sped up through parallelism based on partitioning tables across nodes handling the transactional operations. A first process is performed by a first processor that processes updates for values of a key based on updating a first start time table index using unique keys and a start time field of a row for a first appearance of each unique key from the transactional operations.

Resolving versions in an append-only large-scale data store in distributed data management systems

A method and system of processing graph query are provided. A graph query is received by a relational database graph module. The graph query is translated into one or more relational database queries. One or more relational database queries are translated to be performed on data stored within a relational database. One or more results from the relational database are received based on the sent one or more relational database queries. A synergistic graph is generated on a display, based on the received one or more results.

Supporting synergistic and retrofittable graph queries inside a relational database

According to one or more embodiments of the present invention, a computer-implemented method includes creating a hybrid index for a multi-zone data storage system in which a first zone has data that is organized for transaction processing, and a second zone that has data organized for analytic processing. Each index entry in the hybrid index includes an index key including an equality column, a begin-timestamp column, and a tuple-sequence-number column. The method further includes building an index run to be included in the hybrid index, the index run includes multiple index entries. An index run is created by selecting columns by scanning data blocks from the first zone, computing tuple-sequence-number values for the index entries to be included in the hybrid index, and sorting the index entries according to a predetermined order using the columns to be included in the index run.

Indexing for evolving large-scale datasets in multi-master hybrid transactional and analytical processing systems

Embodiments of the invention relate to executing graph path queries. A database stores data entities and attributes in node tables and stores links between nodes in an edge table. Edges form a path between a source node and a target node. A source node set is generated and joined with the edge table to produce a first intermediate set. Similarly, a target node set is generated and joined with the edge table to produce a second intermediate set. A result path is generated through a joining of the first and second intermediate paths and application of a length condition.

Executing graph path queries

Embodiments relate to executing graph path queries. A database stores data entities and attributes in node tables and stores links between nodes in an edge table. Edges form a path between a source node and a target node. A source node set is generated and joined with the edge table to produce a first intermediate set. Similarly, a target node set is generated and joined with the edge table to produce a second intermediate set. A result path is generated through a joining of the first and second intermediate paths and application of a length condition.

One embodiment provides for a method including processing transactional operations on a key used to determine whether existing data is found for that key. A first time index is updated using unique keys and a start time field of a first appearance of each key from the transactional operations. A deferred update of prior versions of the key is performed for non-recent data upon determining that recent data in the transactional operations is found for the key.

A computer-implemented method, according to one embodiment, includes: generating two or more sample graphs by sampling edges of a current snapshot of a dynamic graph, generating two or more partial results by executing an algorithm on the two or more sample graphs, combining the partial results into a final result, and incrementally maintaining the sample graphs. Edges included in the current snapshot of a dynamic graph and which were added to the dynamic graph in a most recent update thereto are included in each of the generated two or more sample graphs. Moreover, incrementally maintaining the sample graphs includes: subsampling each of the edges of each of the sample graphs at a given time by applying a Bernoulli trial, and combining a result of the subsampling with new edges received in a batch corresponding to the given time to form new sample graphs.

Dynamic interaction graphs with probabilistic edge decay

Parallel execution of machine learning programs is provided. Program code is received. The program code contains at least one parallel for statement having a plurality of iterations. A parallel execution plan is determined for the program code. According to the parallel execution plan, the plurality of iterations is partitioned into a plurality of tasks. Each task comprises at least one iteration. The iterations of each task are independent. Data required by the plurality of tasks is determined. An access pattern by the plurality of tasks of the data is determined. The data is partitioned based on the access pattern.

Hybrid parallelization strategies for machine learning programs on top of mapreduce

In one general embodiment, a computer-implemented method is provided for analyzing a dynamic graph. The computer-implemented method includes generating two or more sample graphs by sampling edges of a current snapshot of a dynamic graph. Additionally, the computer-implemented method includes generating two or more partial results by executing an algorithm on the sample graphs. Still yet, the computer-implemented method includes combining the partial results, from executing the algorithm on the sample graphs, into a final result.

A computer determines social media influencers in a specific topic by receiving a dataset of information associated with a website, the information including a first list of users of the website and a list of content that each user posts on the website, wherein each user is associated with other users from the first list of users. The computer determines initial values representing variables of the dataset of information on the website, wherein the variables include one or more topics for the list of content that each user from the first list of users posts on the website. The computer performs an iteration of Gibbs Sampling utilizing the initial values. The computer determines the one or more new values representing variables of the dataset represent a distribution of the one or more topics for the list of content that each user from the first list of users posts.

Identifying influencers for topics in social media

Embodiments relate to joining data across a parallel database and a distributed processing system. Aspects include receiving a query on data stored in parallel database T and data stored in distributed processing system L, applying local query predicates and projection to data T to create T', and applying local query predicates and projection to L to create L&#x2032;. Based on determining that a size of L&#x2032; is less than a size of T&#x2032; and that the size of L&#x2032; is less than a first threshold, transmitting L&#x2032; to the parallel database and executing a join between T&#x2032; and L&#x2032;. Based on determining that a number of the nodes distributed processing system n multiplied by the size of T&#x2032; is less than the size of L&#x2032; and that the size of T&#x2032; is less than a second threshold; transmitting T&#x2032; to the distributed processing system and executing a join between T&#x2032; and L&#x2032;.

Joining data across a parallel database and a distributed processing system

Growing community of inventors

San Jose, CA, United States of America

Yuanyuan Tian