Temporal Search Research Articles

Temporal JSON Keyword Search

JSON keyword search searches the current versions of documents in a collection. However, JSON documents change over time due to edits. Some applications, such as data forensics and auditing, need to search past versions of documents and for changes to documents. This paper introduces a system called Temporal JSON Keyword Search (TJKS) for search in a collection of JSON documents that vary over time. TJKS lets users control which temporal slice, or part of the history, can be searched using a temporal search semantics; we support both of the major temporal semantics: sequenced and nonsequenced search. This paper presents the semantics of temporal JSON keyword search, discusses an efficient implementation, and evaluates the implementation. Our extensions are largely orthogonal to specific keyword search techniques, so this research provides a blueprint for extending keyword search to include time and potentially other kinds of metadata.

Memento: a prototype search engine for LSC 2021

In this extended paper, we describe our lifelog retrieval system called Memento which participated in the 2021 Lifelog Search Challenge in detail. Memento leverages semantic representations of images and textual queries projected into a common latent space to facilitate effective retrieval, aiming to bridge the existing semantic gap between complex visual scenes/events and user information needs expressed as textual and faceted queries. Our system also has a minimalist user interface which includes functionalities such as visual data filtering and temporal search. Finally, we include a comparative analysis of Memento’s performance at LSC 2021 and suggest improvements for future iterations of the system.

Harvesting BAT-GUANO with NITRATES (Non-Imaging Transient Reconstruction and Temporal Search): Detecting and Localizing the Faintest Gamma-Ray Bursts with a Likelihood Framework

The detection of the gravitational wave (GW) counterpart GRB 170817A, underluminous compared to the cosmological gamma-ray burst (GRB) population by a factor of 10,000, motivates significant effort in detecting and localizing a dim, nearby, and slightly off-axis population of short GRBs. Swift’s Burst Alert Telescope (BAT) is one of the most sensitive GRB detectors in operation, and the only one that regularly localizes GRBs to arcminute precision, critical to rapid follow-up studies. However, the utility of BAT in targeted subthreshold searches had been historically curtailed by the unavailability of the necessary raw data for analysis. The new availability of time-tagged event data from the GUANO system motivates a renewed focus on developing sensitive targeted search analysis techniques to maximally exploit these data. While computationally cheap, we show that the typical coded-mask deconvolution imaging is limited in its sensitivity due to several factors. We formalize a maximum likelihood framework for the analysis of BAT data wherein signals are forward modeled through the full instrument response, and—coupled with the development of new response models—demonstrate its superior sensitivity to typical imaging via archival comparisons, injection campaigns, and a large number of low-latency GRB discoveries and confirmed arcminute localizations to date. We also demonstrate independent localization of some out-of-field-of-view GRBs for the first time. NITRATES’s increased sensitivity boosts the discovery rate of GRB 170817A–like events in BAT by a factor of at least 3−4×, along with enabling joint analyses and searches with other GRB, GW, neutrino, and FRB instruments. We provide public access to the response functions and search pipeline code.

Curricular integration of digital technologies in teaching processes

The integration of digital technologies in the classroom is a complex and multidimensional process with different dynamics including, among others, those related to: the digital culture of the center, the competency of teachers and students, the support of families and innovation within educational programs. This paper presents a systematic literature review (SLR) to analyze how the curricular integration of educational technology in classroom practice has been developed at non-university levels in recent years. The PRISMA 2020 standards have been applied. For the selection of articles, the ERIC database was used, taking as a reference, key concepts from its Thesaurus, related to the objective of the research, performing a temporal search of scientific articles from 2018 up to the present day. After screening according to the inclusion criteria established by consensus among researchers, a total of 88 articles were obtained (n = 88). The main results point to several variables that should be strengthened to promote the integration of digital technologies in the classroom, among which teacher training stands out as a determining factor, with special emphasis on initial training. This opens a debate about the training that future teachers receive in relation to the integration of digital technologies in the teaching process.Systematic review registration[https://doi.org/10.5281/zenodo.6909261].

Amitriptyline: Useful for the prophylactic treatment of migraine?

Migraine is a type of headache described by recurrent attacks that may be accompanied by some symptoms, such as nausea, vomiting, phonophobia, among other conditions. About 1 billion people in the world could come into existence. Some measurements are recognized for the treatment of use, as measurements are made appropriately with cost drugs. Amitriptyline has become a great alternative for the prophylactic treatment of migraine, its metabolization is extensively in the liver and its analgesic effect is linked to the ability to modulate mood or modulate the excitability of neurons that transmit pain signals from the dura mater. This study is an integrative literature review. A search key was created, and the search was performed in the Web of Science (WoS-all databases). A temporal search was carried out between the years 2018 to June 2022, without language restriction. Tools such as MapChart were used to create the worldwide distribution map based on the corresponding author. 100 articles were rescued from the Web of Science, of which only 20 were chosen to compose the final table, due to their high relevance and sensitivity in the investigated topic. Migraine prevention for patients is critical, as it can reduce stigma and enhance quality of life.

Temporal search persistence, certainty, and source preference in dentistry: Results from the National Dental PBRN

The primary goal of this paper was to investigate an old question in a new way: what are the search patterns that professionals demonstrate when faced with a specific knowledge gap? We examine data from a cascading survey question design that captures details about searching for information to answer a self-nominated clinical question from 1027 dental professionals enrolled in the National Dental Practice Based Research Network. Descriptive and conditional logistical regression analysis techniques were used. 61% of professionals in our sample choose informal sources of information, with only about 11% looking to formal peer reviewed evidence. The numbers of professionals turning to general internet searches is more than twice as high as any other information source other than professional colleagues. Dentists with advanced training and specialists are significantly more likely to consult peer-reviewed sources, and women in the sample were more likely than men to continue searching past a first source. Speed/availability of information may be just as, or in some cases, more important than credibility for professionals' search behavior. Additionally, our findings suggest that more insights are needed into how various categories of professionals within a profession seek information differently.

Adjustments of selective attention to response conflict \u2013 controlling for perceptual conflict, target-distractor identity, and congruency level sequence pertaining to the congruency sequence effect

The congruency sequence effect (CSE) describes the performance difference of congruent trials (in which target and distractor stimuli are associated with the same response) compared to incongruent trials (in which target and distractor stimuli are associated with different responses) as a function of the preceding congruency level (congruent trials relative to incongruent trials). The CSE is commonly interpreted as a measure of conflict-induced attentional adjustment. Although previous research has made substantial progress aiming at controlling for alternative explanations of the CSE, both task-specific and fundamental confounds have remained. In the current study, we used a temporal flanker task, in which two stimuli (i.e., distractor and target) are presented in rapid succession, and extended previous demonstrations of a CSE in flanker tasks by deconfounding target-distractor congruency from perceptual similarity. Using a four-choice task, we could also control for the reversal of distractor-response priming after incongruent trials (which is only feasible in two-choice tasks). Furthermore, we controlled for all confounds based on the sequence (i.e., repetition versus alternation) of the congruency level – such as feature sequence effects, distractor-response contingency switch costs, or temporal learning – by probing the allocation of attention to the points in time of presentation of the first and the second stimulus of a trial. This was achieved by intermixing trials of a temporal search task. The performance accuracy results in this task were consistent with a stronger attentional bias in favor of the target stimulus’ temporal position after incongruent than after congruent trials.

Assessing Public Interest Based on Wikipedia's Most Visited Medical Articles During the SARS-CoV-2 Outbreak: Search Trends Analysis.

BackgroundIn the current era of widespread access to the internet, we can monitor public interest in a topic via information-targeted web browsing. We sought to provide direct proof of the global population’s altered use of Wikipedia medical knowledge resulting from the new COVID-19 pandemic and related global restrictions.ObjectiveWe aimed to identify temporal search trends and quantify changes in access to Wikipedia Medicine Project articles that were related to the COVID-19 pandemic.MethodsWe performed a retrospective analysis of medical articles across nine language versions of Wikipedia and country-specific statistics for registered COVID-19 deaths. The observed patterns were compared to a forecast model of Wikipedia use, which was trained on data from 2015 to 2019. The model comprehensively analyzed specific articles and similarities between access count data from before (ie, several years prior) and during the COVID-19 pandemic. Wikipedia articles that were linked to those directly associated with the pandemic were evaluated in terms of degrees of separation and analyzed to identify similarities in access counts. We assessed the correlation between article access counts and the number of diagnosed COVID-19 cases and deaths to identify factors that drove interest in these articles and shifts in public interest during the subsequent phases of the pandemic.ResultsWe observed a significant (P<.001) increase in the number of entries on Wikipedia medical articles during the pandemic period. The increased interest in COVID-19–related articles temporally correlated with the number of global COVID-19 deaths and consistently correlated with the number of region-specific COVID-19 deaths. Articles with low degrees of separation were significantly similar (P<.001) in terms of access patterns that were indicative of information-seeking patterns.ConclusionsThe analysis of Wikipedia medical article popularity could be a viable method for epidemiologic surveillance, as it provides important information about the reasons behind public attention and factors that sustain public interest in the long term. Moreover, Wikipedia users can potentially be directed to credible and valuable information sources that are linked with the most prominent articles.

ACE: the Advanced Cohort Engine for searching longitudinal patient records.

ObjectiveTo propose a paradigm for a scalable time-aware clinical data search, and to describe the design, implementation and use of a search engine realizing this paradigm.Materials and MethodsThe Advanced Cohort Engine (ACE) uses a temporal query language and in-memory datastore of patient objects to provide a fast, scalable, and expressive time-aware search. ACE accepts data in the Observational Medicine Outcomes Partnership Common Data Model, and is configurable to balance performance with compute cost. ACE’s temporal query language supports automatic query expansion using clinical knowledge graphs. The ACE API can be used with R, Python, Java, HTTP, and a Web UI.ResultsACE offers an expressive query language for complex temporal search across many clinical data types with multiple output options. ACE enables electronic phenotyping and cohort-building with subsecond response times in searching the data of millions of patients for a variety of use cases.DiscussionACE enables fast, time-aware search using a patient object-centric datastore, thereby overcoming many technical and design shortcomings of relational algebra-based querying. Integrating electronic phenotype development with cohort-building enables a variety of high-value uses for a learning health system. Tradeoffs include the need to learn a new query language and the technical setup burden.ConclusionACE is a tool that combines a unique query language for time-aware search of longitudinal patient records with a patient object datastore for rapid electronic phenotyping, cohort extraction, and exploratory data analyses.

Development of a Method for Determining the Search Window for Solar Flare Neutrinos

Neutrinos generated during solar flares remain elusive. However, after $50$ years of discussion and search, the potential knowledge unleashed by their discovery keeps the search crucial. Neutrinos associated with solar flares provide information on otherwise poorly known particle acceleration mechanisms during solar flare. For neutrino detectors, the separation between atmospheric neutrinos and solar flare neutrinos is technically encumbered by an energy band overlap. To improve differentiation from background neutrinos, we developed a method to determine the temporal search window for neutrino production during solar flares. Our method is based on data recorded by solar satellites, such as Geostationary Operational Environmental Satellite (GOES), Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI), and GEOTAIL. In this study, we selected 23 solar flares above the X5.0 class that occurred between 1996 and 2018. We analyzed the light curves of soft X-rays, hard X-rays, $\gamma$-rays, line $\gamma$-rays from neutron capture as well as the derivative of soft X-rays. The average search windows are determined as follows: $4,178$ s for soft X-ray, $700$ s for derivative of soft X-ray, $944$ s for hard X-ray ($100$-$800$ keV), $1,586$ s for line $\gamma$-ray from neutron captures, and $776$ s for hard X-ray (above $50$ keV). This method allows neutrino detectors to improve their sensitivity to solar flare neutrinos.

A systematic review of internet-based information on dermatomyositis and polymyositis.

Patients with rheumatic diseases are increasingly using internet-based information to inform healthcare utilization and make treatment decisions. Our aim was to assess the readability and quality of internet-based information on dermatomyositis (DM) and polymyositis (PM). Key words "Dermatomyositis" and "Polymyositis" were searched on 3 commonly used search engines (Google, Yahoo and Bing). The first 3 pages (~30) of search results were examined from each search engine. Readability of information was assessed using 4 readability formulae (Flesch Reading Ease Score, Flesch-Kincaid Grade Level, the Simplified Measure of Gobbledygook index, the Coleman-Liau index). Quality of information was assessed using the DISCERN tool, Journal of The American Medical Association (JAMA) benchmark criteria and Health on The Net Code (HoN code). We also examined Google Trends® data to determine if there were obvious temporal search patterns. Thirty-two websites were included in the study after duplicates were removed and exclusion criteria were applied. The overall quality was low including DISCERN with a median overall score of 38/80 (interquartile range 12.25), only 4/32 (13%) websites fulfilled all 4 JAMA benchmark criteria, and 9/32 (28%) had HoN code. Readability of information was assessed using 4 readability formulae (Flesch Reading Ease Score, Flesch-Kincaid Grade Level, the Simplified Measure of Gobbledygook index, the Coleman-Liau index. There was no obvious temporal trend in searches on analysis of Google Trends® data. The overall quality and readability of internet-based information relating to DM and PM is poor. Patients require appropriate information of high quality and readability throughout the course of their disease in order to make informed decisions on their condition including treatment.

Search and concealment strategies in the spatiotemporal domain.

Although visual search studies have primarily focused on search behavior, concealment behavior is also important in the real world. However, previous studies in this regard are limited in that their findings about search and concealment strategies are restricted to the spatial (two-dimensional) domain. Thus, this study evaluated strategies during three-dimensional and temporal (i.e., spatiotemporal) search and concealment to determine whether participants would indicate where they would hide or find a target in a temporal sequence of items. The items were stacked in an upward (Experiments 1-3) or downward (Experiment 4) direction and three factors were manipulated: scenario (hide vs. seek), partner type (friend vs. foe), and oddball (unique item in the sequence; present vs. absent). Participants in both the hide and seek scenarios frequently selected the oddball for friends but not foes, which suggests that they applied common strategies because the oddball automatically attracts attention and can be readily discovered by friends. Additionally, a principle unique to the spatiotemporal domain was revealed, i.e., when the oddball was absent, participants in both scenarios frequently selected the topmost item of the stacked layer for friends, regardless of temporal order, whereas they selected the first item in the sequence for foes, regardless of the stacked direction. These principles were not affected by visual masking or number of items in the sequence. Taken together, these results suggest that finding and hiding positions in the spatiotemporal domain rely on the presence of salient items and physical accessibility or temporal remoteness, according to partner type.

GeoDataHub: Building centralised repository of authoritative cartographic and geospatial resources

Abstract. GeoDataHub (geodatahub.library.auckland.ac.nz) is an authoritative national repository of cartographic and geospatial resources curated by the University of Auckland Library for research, government, and educational communities, as well as the public of New Zealand. Its objectives are to digitise cartographic materials, provide integrated and enhanced data access, and build a unique knowledge base concerning New Zealand’s cartographic heritage. GeoDataHub showcases conceptual developments in cartography, surveying, photogrammetry, and other mapping, earth, and social science disciplines, highlighting their contribution to the development of the nation. It supports a dynamic, innovative, and leading edge historical and GIScience research, facilitating scientific discovery and knowledge building. The repository contains digitised maps and charts published by authoritative agencies and research institutes, aerial photography provided by local authorities and all of government satellite imagery. A selection of key geospatial datasets produced by the government, Crown research institutes and local authorities augment the repository – currently over 30TB of data. The focus of GeoDataHub is on providing cartographic and geospatial resources for New Zealand, its offshore islands and territories; South Pacific, particularly Cook Islands, Niue and Tokelau; and Ross Dependency in Antarctica. Regional coverage primarily includes Auckland and Waikato, and parts of Northland, Bay of Plenty and East Cape. Local datasets cover primarily Auckland Region but a selection of datasets from other regions is also included. GeoDataHub is augmented by physical collection of maps and charts, atlases, aerial photographs, gazetteers, map indexes and catalogues, and cartographic paraphernalia, located in the Map Room of the University of Auckland Library. This paper talks about challenges associated with the design, implementation and provision of content for GeoDataHub. Apart from building an integrated and enhanced data access application, facilitating textual, spatial and temporal search, metadata harvesting and discovery utilising GeoNetwork, a catalogue application that manages spatially referenced resources (www.geonetwork-opensource.org), the repository provides a further two ways of accessing cartographic and geospatial resources. One involves browsing the selected content of the repository, examining and downloading the material of interest; the other allows users to ‘map a drive’ and access all materials using familiar folder structure (the latter applies to the University community of users only). This approach however, occasionally requires provision of the same datasets in two different formats (compressed and uncompressed) depending how the users are accessing them. In addition, a knowledge base concerning New Zealand cartographic heritage is being implemented, providing a useful guide to authoritative mapping served by GeoDataHub. A particular challenge concerns documenting their provenance, content, extent and general characteristics in a modular fasion. The paper concludes with presenting plans for extending GeoDataHub. This includes not only the creation of further geospatial records, provision of more content and expanding the knowledge base, but also strategic decisions concerning the selection of cartographic and geospatial resources and how to provide the best service possible to our growing user base.

Sex Differences in Temporal but Not Spatial Attentional Capture.

The accuracy of detecting or identifying a target decreases when a salient distractor is presented. This decrease is explained by the temporal or spatial diversion of attention to the distractor and thus is referred to as attentional capture. Using temporal and spatial visual search tasks, we examined whether there are sex differences in attentional capture. In Experiment 1A, a temporal visual search task measured attentional capture in the temporal domain by asking participants (97 men and 92 women) to identify a target embedded in a rapid stream of nontarget letters while ignoring a preceding peripheral distractor. In Experiment 2, a spatial visual search task measured attentional capture in the spatial domain by asking participants (146 men and 83 women) to detect a target among spatially distributed nontarget items while ignoring a distractor presented simultaneously. Our results indicate that attentional capture occurred in both tasks. In Experiment 1A, the magnitude of capture was significantly larger for women than men. In Experiment 1B, we confirmed sex differences in temporal attentional capture by recruiting a new set of participants (141 men and 85 women). In Experiment 2, the magnitude of capture was comparable between the sexes. These results suggest that women are more sensitive to bottom-up signals than men when they engage in a temporal search task and could be explained in terms of sex differences in the ability of adjusting the size of attentional window, within which attention is allocated to the most salient item.

Target-distractor congruency: sequential effects in a temporal flanker task.

The Congruency Sequence Effect (CSE) denotes the common finding that distractor-target Congruency Effects are reduced after incongruent compared to after congruent trials. Although the CSE is widely attributed to attentional adjustment (i.e., increasing or decreasing the bias in attentional weights regarding processing the target or distractor), unequivocal evidence for this assumption is missing. To investigate the CSE and attentional adjustment we used a temporal flanker task and intermixed a "temporal search task", in which a target stimulus occurred randomly at one of two temporal positions, corresponding to the temporal positions of the target and the distractor occurrence in the temporal flanker task. We observed a CSE that could not be explained in terms of feature sequences, distractor-related contingencies, or a strategy of reversed distractor-response priming after incongruent trials. Furthermore, following a temporal search task trial, the Congruency Effect was larger when the search target occurred on the first than on the second temporal position, demonstrating that a reduced attentional bias towards the second temporal position increased interference from a distractor presented on the first temporal position. This supports a crucial assumption of the attentional adjustment account of the CSE. Performance in the temporal search task, however, provided no evidence for attentional adjustment.

Abstract B71: New directions in gynecologic cancer research utilizing Text Information Extraction System (TIES) Cancer Research Network

Abstract Gynecologic cancers pose significant morbidity and mortality burden for US women. Endometrial cancer is the most common gynecologic malignancy in the US, while ovarian cancer is one of the deadliest. Since research funding for some of the gynecologic malignancies has been limited, novel research on the existing datasets needs to be harnessed to improve the health of the women affected by these conditions. We aim to evaluate if Text Information Extraction System (TIES) Cancer Research Network, a federated network that facilitates data and biospecimen sharing among member institutions, has capabilities for conducting research in the area of gynecologic cancers. TIES system provides effective tools for searching deidentified structured and unstructured data for gynecologic cancers with capabilities to explore pathology and radiology records. We used TIES system to identify records of patients with ovarian, endometrial, and cervical cancers. We used temporal search features to identify patients who developed ovarian tumors after being diagnosed with endometrial tumors. We examined records for completeness of information on diagnosis, age, and procedure year. We found that TIES system has 13,762 records for endometrial cancer, 7,692 records for ovarian cancer, and 10,826 records for cervical cancer for Pittsburgh site. 306 records have been identified for patients who had both ovarian and endometrial malignancies. Nearly 100% of the records had information on age of the patient, year of the procedure, and exact diagnosis. TIES is a promising tool for conducting research in the area of gynecologic malignancies, especially for rare diseases such as concurrent endometrial and ovarian cancers. Existing data repositories like TIES need to be further explored for novel research questions in the area of gynecologic cancer. Citation Format: Faina Y. Linkov, Melissa Schwenk, Michael Davis, Robert Edwards, Michael Becich. New directions in gynecologic cancer research utilizing Text Information Extraction System (TIES) Cancer Research Network. [abstract]. In: Proceedings of the AACR Conference: Addressing Critical Questions in Ovarian Cancer Research and Treatment; Oct 1-4, 2017; Pittsburgh, PA. Philadelphia (PA): AACR; Clin Cancer Res 2018;24(15_Suppl):Abstract nr B71.

MAKING TEMPORAL SEARCH MORE CENTRAL IN SPATIAL DATA INFRASTRUCTURES

Abstract. A temporally enabled Spatial Data Infrastructure (SDI) is a framework of geospatial data, metadata, users, and tools intended to provide an efficient and flexible way to use spatial information which includes the historical dimension. One of the key software components of an SDI is the catalogue service which is needed to discover, query, and manage the metadata. A search engine is a software system capable of supporting fast and reliable search, which may use any means necessary to get users to the resources they need quickly and efficiently. These techniques may include features such as full text search, natural language processing, weighted results, temporal search based on enrichment, visualization of patterns in distributions of results in time and space using temporal and spatial faceting, and many others. In this paper we will focus on the temporal aspects of search which include temporal enrichment using a time miner - a software engine able to search for date components within a larger block of text, the storage of time ranges in the search engine, handling historical dates, and the use of temporal histograms in the user interface to display the temporal distribution of search results.

Efficient rate control scheme using modified inter-layer dependency for spatial scalability

Scalable video coding extension has been added to H.264AVC to support compression and encoding of multiple resolution video sequences, having different frame rates and fidelities in a single bit stream. The motion vectors and the residual data of the enhancement layers are derived from up-sampling the co-located macroblock (MB) of the base layer. The peak signal to noise ratio (PSNR) across the enhancement layers is degraded as up-sampling introduces distortion of high-frequency components. In this paper, a spatial-resolution-ratio-based MB mode decision scheme is proposed for spatially enhanced layers. The scheme uses the motion estimated at the base layer, to encode the respective MBs in the enhancement layers. The spatial–temporal search schemes at the enhancement layers are used to derive motion vectors and residues that are encoded using a quantization parameter obtained using independent rate control (IRC) scheme. The IRC from the prior art is modified to achieve better rate control per layer by recursive updates for mean absolute difference values of each basic unit. Proposed modified inter-layer dependency shows improvement in the PSNR for enhancement layers while the updated IRC enforces better IRC for all the layers.

When time meets information retrieval: Past proposals, current plans and future trends

With the advent of Web search and the large amount of data published on the Web sphere, a tremendous amount of documents become strongly time-dependent. In this respect, the time dimension has been extensively exploited as a highly important relevance criterion to improve the retrieval effectiveness of document ranking models. Thus, a compelling research interest is going on the temporal information retrieval realm, which gives rise to several temporal search applications. In this article, we intend to provide a scrutinizing overview of time-aware information retrieval models. We specifically put the focus on the use of timeliness and its impact on the global value of relevance as well as on the retrieval effectiveness. First, we attempt to motivate the importance of temporal signals, whenever combined with other relevance features, in accounting for document relevance. Then, we review the relevant studies standing at the crossroads of both information retrieval and time according to three common information retrieval aspects: the query level, the document content level and the document ranking model level. We organize the related temporal-based approaches around specific information retrieval tasks and regarding the task at hand, we emphasize the importance of results presentation and particularly timelines to the end user. We also report a set of relevant research trends and avenues that can be explored in the future.

Object Instance Search in Videos via Spatio-Temporal Trajectory Discovery

Given a specific object as query, object instance search aims to not only retrieve the images or frames that contain the query, but also locate all its occurrences. In this work, we explore the use of spatio-temporal cues to improve the quality of object instance search from videos. To this end, we formulate this problem as the spatio-temporal trajectory search problem, where a trajectory is a sequence of bounding boxes that locate the object instance in each frame. The goal is to find the top- $K$ trajectories that are likely to contain the target object. Despite the large number of trajectory candidates, we build on a recent spatio- temporal search algorithm for event detection to efficiently find the optimal spatio- temporal trajectories in large video volumes , with complexity linear to the video volume size. We solve the key bottleneck in applying this approach to object instance search by leveraging a randomized approach to enable fast scoring of any bounding boxes in the video volume. In addition , we present a new dataset for video object instance search. Experimental results on a 73-hour video dataset demonstrate that our approach improves the performance of video object instance search and localization over the state-of-the-art search and tracking methods.