In this paper, we study the problem of numerical multi-table question answering (MTQA) over large-scale table collections (e.g., online data repositories). This task is essential in many analytical applications. Existing MTQA solutions, such as text-to-SQL or open-domain MTQA methods, are designed for databases and struggle when applied to large-scale table collections. The key limitations include: (1) Limited support for complex table relationships; (2) Ineffective retrieval of relevant tables at scale; (3) Inaccurate answer generation. To overcome these limitations, we propose DMRAL, a Decomposition-driven Multi-table Retrieval and Answering framework for MTQA over large-scale table collections, which consists of: (1) constructing a table relationship graph to capture complex relationships among tables; (2) Table-Aligned Question Decomposer and Coverage-Aware Retriever, which jointly enable the effective identification of relevant tables from large-scale corpora by enhancing the question decomposition quality and maximizing the question coverage of retrieved tables; and (3) Sub-question Guided Reasoner, which produces correct answers by progressively generating and refining the reasoning program based on sub-questions. Experiments on two MTQA datasets demonstrate that DMRAL significantly outperforms existing state-of-the-art MTQA methods, with an average improvement of 24% in table retrieval and 55% in answer accuracy.

Abstract Missing values in tabular data lakes can severely impact data analysis and diminish the performance in downstream applications. We highlight that a robust imputation strategy should properly take three aspects of variety into consideration: source of imputed value, the types of tables involved, and the data types of the missing value. Existing imputation methods rely on estimation-based approaches (using a model trained on data from the same table to estimate missing values) or search-based approaches (retrieving values from other tables). Unfortunately, none of these approaches effectively incorporate all three aspects of variety. To address this gap, we propose , a novel framework that uses a C ombination of E stimation-based and S earch-based methods for missing value I mputation in D ata lakes. contains three core modules: (1) the , which efficiently discovers candidate values from tables by exploiting the contextual information; (2) the , which introduces an influence function and a sampling-based exploration strategy to yield accurate estimated values; (3) the , which determines the most suitable method based on table-level and column-level statistics. Extensive experiments conducted on three data lakes demonstrate that effectively and efficiently addresses the missing value problem.

This paper proposes a novel graph structure to address the problems of information spreading in a real-world, frequently updating graph, with two main contributions at hand: accurately tracing infection diffusion according to fine-grained user movements and finding vulnerable vertices under the virus immunization scenario to mitigate infection diffusion. Unlike previous work that primarily predicts the long-term epidemic trend at the census level, this study aims to intervene in the short-term at the individual level. Therefore, two downstream tasks are formulated to illustrate practicalities:EpidemicMitigating in PublicArea problem ($EMA$) andEpidemic MaximizedSpread in PublicArea problem ($ESA$), where$EMA$aims to find intervention strategies, and$ESA$is an adversarial solution against the intervention strategy to test the robustness. Comprehensive experiments are conducted using two real-world datasets with millions of public transport trips, which demonstrate the effectiveness of our approach and highlight the importance of considering the dynamic nature of close contacts in epidemic modelling.

Abstract Patient similarity search is an essential task in healthcare. Recent studies adopted electronic health records (EHRs) to learn patient representations for measuring the clinical similarities. These methods outperformed traditional methods, by capturing more information from various sources consisting of multi-modal EHRs, external knowledge and correlations among medical concepts. They often concerned certain type of data without taking full advantage of various information. We propose a graph representation learning framework, denoted by One-Size-Fits-Three ( OSFT ), that takes into account fusion-attention, neighbor-attention and global-attention from three types of information. Extensive experiments are conducted on two real datasets of MIMIC-III and MIMIC-IV, and the results verified the effectiveness and generality of our framework. When compared with baselines on patient similarity search, our framework achieved good effectiveness and comparative efficiency. The results provide new insights about whether the use of various information can better measure the patient similarity. The source codes are available at https://github.com/emmali808/ADDS/tree/master/EHRDeepHelper .

Healthcare question answering (HQA) is a challenging task as questions are generally non-factoid. Recent neural systems are reported to have performance gains. However, little attention has been given to HQA as datasets are generally too small to train a neural model from scratch. Recently, several systems have been proposed to learn context representations for HQA. Despite moderate progress, these systems have not been thoroughly compared with state-of-the-art neural models, and the mentioned models are tested only on self-created datasets. To address the challenges, we develop a new joint model to incorporate both context and knowledge embeddings into neural ranking architectures. First, we adapt context embedding pre-trained from large open-domain corpus to small healthcare datasets. Second, we learn knowledge embedding from knowledge graphs to provide external information for understanding non-factoid questions. To evaluate our framework, we adapt many state-of-the-art methods for general QA to HQA, by injecting the context or knowledge information only, or both of them. Extensive experiments are conducted to compare our approach with those adapted methods and current HQA systems. The results show that our approach achieves the state-of-the-art performance on both HealthQA and NFCorpus datasets. The source codes are available at https://github.com/emmali808/HQADeepHelper.

Medical visual question answering (Med-VQA) is a challenging problem that aims to take a medical image and a clinical question about the image as input and output a correct answer in natural language. Current medical systems often require large-scale and high-quality labeled data for training and evaluation. To address the challenge, we present a new dataset, denoted by OVQA, which is generated from electronic medical records. We develop a semi-automatic data generation tool for constructing the dataset. First, medical entities are automatically extracted from medical records and filled into predefined templates for generating question and answer pairs. These pairs are then combined with medical images extracted from corresponding medical records, to generate candidates for visual question answering (VQA). The candidates are finally verified with high-quality labels annotated by experienced physicians. To evaluate the quality of OVQA, we conduct comprehensive experiments on state-of-the-art methods for the Med-VQA task to our dataset. The results show that our OVQA can be used as a benchmarking dataset for evaluating existing Med-VQA systems. The dataset can be downloaded from http://47.94.174.82/.

Entity resolution (ER) identifies data instances referring to the same real-world entity and has received enormous research attention. In this paper, we examine the task of ER from a broader perspective, with its input extended from textual records, which are conventionally studied in the literature, to other modalities such as check-in sequences, GPS trajectories and surveillance video frames to generate new applications. Our goal in this paper is to design an effective model to uniformly support all these ER applications with different input formats. Technically, we fully exploit the semantic contexts of embedding vectors for the pair of input instances. In particular, we propose an integrated multi-context attention framework that takes into account self-attention, pair-attention and global-attention from three types of context. The idea can be further extended to incorporate attribute attention in order to support structured datasets. We conduct extensive experiments on a diverse class of entity resolutions tasks, including tasks on unstructured, structured and dirty textual records, check-in sequences, GPS trajectories and surveillance video frames. The experimental results verified the effectiveness and generality of our model. When compared with strong baselines in these applications, our model can achieve superior or comparative performance.

This paper demonstrates a medical visual question answering (VQA) system to address three challenges: 1) medical VQA often lacks large-scale labeled training data which requires huge efforts to build; 2) it is costly to implement and thoroughly compare medical VQA models on self-created datasets; 3) applying general VQA models to the medical domain by transfer learning is challenging due to various visual concepts between general images and medical images. Our system has three main components: data generation, model library, and model practice. To address the first challenge, we first allow users to upload self-collected clinical data such as electronic medical records (EMRs) to the data generation component and provides an annotating tool for labeling the data. Then, the system semi-automatically generates medical VQAs for users. Second, we develop a model library by implementing VQA models for users to evaluate their datasets. Users can do simple configurations by selecting self-interested models. The system then automatically trains the models, conducts extensive experimental evaluation, and reports comprehensive findings. The reports provide new insights into the strengths and weaknesses of selected models. Third, we provide an online chat module for users to communicate with an AI robots for further evaluating the models. The source codes are shared on https://github.com/shyanneshan/VQA-Demo.

This paper demonstrates an intelligent medical analysis system. We aim to address two main challenges: 1) medical data often contain heterogeneous information which are usually valuable but difficult to be modeled; 2) medical data are often lacking of large scale labeled data which usually require huge efforts to build. To resolve the first challenge, we propose a novel multi-modal heterogeneous graph model to represent the medical data. Based on this model, graph neural networks can be directly applied to effective medical case clustering. This helps to resolve the second challenge for label assignment in the same cluster. To further evaluate the practical use of the proposed model, the system also proposes an effective similar medical case retrieval framework based on a novel graph similarity learning model. We have implemented the system and the source codes are published at https://github.com/emmali808/ADDS. With our system, users can easily pinpoint valuable historical medical information they are interested in and obtain closely relevant medical cases for further diagnosis.

Description of Points of Interest (POIs) plays an important role to enhance the quality of many location-based services, such as displaying concentrated information of POIs for user-friendly experience and leading to successful POI recommendation. However, only a few popular POIs have enough description on the web. Collecting or writing high-quality descriptions for many unpopular or long-tail POIs remains a huge challenge for online map services, especially considering there are numerous new appeared POIs every day. Unlike existing studies about automatic product description generation, the POI description is quite diverse across different locations over a country, and requires high expert knowledge. To address this issue, we first study the POI description generation problem by proposing a novel model, named as Multi Mode Description Generator (MMDG), to automatically generate description based on POIs' reviews and other features. To extract key information for POI description generation, MMDG is equipped with a multi-mode encoder and a transformer-based decoder. Besides user reviews, the multi-mode encoder also considers the category and spatial context information of target POIs, and integrate them with a fusion function. We have conducted an extensive experimental evaluation on a large-scale real-world dataset to demonstrate its effectiveness and superiority over state-of-the-art baselines in terms of various metrics.

Online text streams such as Twitter are the major information source for users when they are looking for ongoing events. Realtime event summarization aims to generate and update coherent and concise summaries to describe the state of a given event. Due to the enormous volume of continuously coming texts, realtime event summarization has become the de facto tool to facilitate information acquisition. However, there exists a challenging yet unexplored issue in current text summarization techniques: how to preserve the integrity, i.e., the accuracy and consistency of summaries during the update process. The issue is critical since online text stream is dynamic and conflicting information could spread during the event period. For example, conflicting numbers of death and injuries might be reported after an earthquake. Such misleading information should not appear in the earthquake summary at any timestamp. In this article, we present a novel realtime event summarization framework called IAEA (i.e., Integrity-Aware Extractive-Abstractive realtime event summarization). Our key idea is to integrate an inconsistency detection module into a unified extractive–abstractive framework. In each update, important new tweets are first extracted in an extractive module, and the extraction is refined by explicitly detecting inconsistency between new tweets and previous summaries. The extractive module is able to capture the sentence-level attention which is later used by an abstractive module to obtain the word-level attention. Finally, the word-level attention is leveraged to rephrase words. We conduct comprehensive experiments on real-world datasets. To reduce efforts required for building sufficient training data, we also provide automatic labeling steps of which the effectiveness has been empirically verified. Through experiments, we demonstrate that IAEA can generate better summaries with consistent information than state-of-the-art approaches.

Mobile devices have generated massive textual data with geographical locations. These data are applied to the topic discovery, event detection and user behavior analysis in enterprise systems. Therefore, geospatial data mining has become a very important and challenging research topic in such systems. In this paper, we develop a geospatial data mining system called GDMS to support the retrieval and analysis of textual data with geographical locations. The system contains three components: data collection, data analysis and data visualization. First, a large number of geospatical data are collected from our implemented mobile APP that is used by community residents. Residents can use the APP to upload abnormal events by text descriptions with geographical locations. All these events are processed and stored in a server. In the data analysis component, we focus on the problem of finding textual topics of clusters containing text descriptions with geographical locations. The key is how to combine clustering techniques with topic-retrieval models to integrate both geo-location information and text information. We investigated methods that combine clustering methods with the knowledge graph to discover topics of clusters of documents with geo-locations. Finally, we demonstrate an effective visualization tool that shows detected textual topics on the map in our mobile APP that is used by government staffs.

Medical archives processing is a very important task in a medical information system. It generally consists of three steps: medical archives recognition, feature extraction and text classification. In this paper, we focus on empowering the medical archives processing with knowledge graphs. We first build a semantic-rich medical knowledge graph. Then, we recognize texts from medical archives using several popular optical character recognition (OCR) engines, and extract keywords from texts using a knowledge graph based feature extraction algorithm. Third, we define a semantic measure based on knowledge graph to evaluate the similarity between medical texts, and perform the text classification task. This measure can value semantic relatedness between medical documents, to enhance the text classification. We use medical archives collected from real hospitals for validation. The results show that our algorithms can significantly outperform typical baselines that employs only term statistics.

With rapid development of Q&A sites such as Quora and StackExchange, high quality question-answer pairs have been produced by users. These Q&A contents cover a wide range of topics, and they are useful for users to resolve queries and obtain new knowledge. Meanwhile, when people are reading digital documents, they may encounter reading problems such as lack of background information and unclear illustration of concepts. We believe that Q&A sites offer high-quality contents which can serve as rich supplements to digital documents. In this paper, we devise a rigorous formulation of the novel text enrichment problem, and design an end-to-end system named QALink which assigns the most relevant Q&A contents to the corresponding section of the document. We first present a new segmentation approach to model each document with a hierarchical structure. Based on the hierarchy, queries are constructed to retrieve and rank related question-answer pairs. Both syntactical and semantic features are adopted in our system. The empirical evaluation results indicate that QALink is able to effectively enrich text documents with relevant Q&A contents to help people better understand the documents.

We demonstrate a system, DocRicher, to enrich a text document with social media, that implicitly reference certain passages of it. The aim is to provide an automatic annotation interface to satisfy users' information need, without cumbersome queries to traditional search engines. The system consists of four components: text analysis, query construction, data assignment, and user feedback. Through text analysis, the system decomposes a text document into appropriate topical passages, of which each is represented using detected key phrases. By submitting combinations of these phrases as queries to social media systems, the relevant results are used to suggest new annotations, that are linked to the corresponding passages. We have built a user-friendly visualization tool for users to browse automatically recommended annotations on their reading documents. Users are either allowed to rate a recommended annotation by accepting it or not; or add a new annotation by manually highlighting texts and adding personal comments. Both these annotations are regarded as the ground truth to derive new queries for retrieving more relevant contents. We also apply data fusion to merge the query results from various contexts and retain most relevant ones.

In crowdsourcing database, human operators are embedded into the database engine and collaborate with other conventional database operators to process the queries. Each human operator publishes small HITs (Human Intelligent Task) to the crowdsourcing platform, which consists of a set of database records and corresponding questions for human workers. The human workers complete the HITs and return the results to the crowdsourcing database for further processing. In practice, published records in HITs may contain sensitive attributes, probably causing privacy leakage so that malicious workers could link them with other public databases to reveal individual private information. Conventional privacy protection techniques, such asK-Anonymity, can be applied to partially solve the problem. However, after generalizing the data, the result of standard K-Anonymity algorithms may render uncontrollable information loss and affects the accuracy of crowdsourcing. In this paper, we first study the tradeoff between the privacy and accuracy for the human operator within data anonymization process. A probability model is proposed to estimate the lower bound and upper bound of the accuracy for general K-Anonymity approaches. We show that searching the optimal anonymity approach is NP-Hard and only heuristic approach is available. The second contribution of the paper is a general feedback-based K-Anonymity scheme. In our scheme, synthetic samples are published to the human workers, the results of which are used to guide the selection on anonymity strategies. We apply the scheme on Mondrian algorithm by adaptively cutting the dimensions based on our feedback results on the synthetic samples. We evaluate the performance of the feedback-based approach on U.S. census dataset, and show that given a predefined$K$, our proposal outperforms standard K-Anonymity approaches on retaining the effectiveness of crowdsourcing.

In this paper, we address the problem of finding k -nearest neighbors (KNN) in sequence databases using the edit distance. Unlike most existing works using short and exact n -gram matchings together with a filter-and-refine framework for KNN sequence search, our new approach allows us to use longer but approximate n -gram matchings as a basis of KNN candidates pruning. Based on this new idea, we devise a pipeline framework over a two-level index for searching KNN in the sequence database. By coupling this framework together with several efficient filtering strategies, i.e. the frequency queue and the well-known Combined Algorithm (CA), our proposal brings various enticing advantages over existing works, including 1) huge reduction on false positive candidates to avoid large overheads on candidate verifications; 2) progressive result update and early termination; and 3) good extensibility to parallel computation. We conduct extensive experiments on three real datasets to verify the superiority of the proposed framework.

The string similarity join is a basic operation of many applications that need to find all string pairs from a collection given a similarity function and a user-specified threshold. Recently, there has been considerable interest in designing new algorithms with the assistant of an inverted index to support efficient string similarity joins. These algorithms typically adopt a two-step filter-and-refine approach in identifying similar string pairs: 1) generating candidate pairs by traversing the inverted index; and 2) verifying the candidate pairs by computing the similarity. However, these algorithms either suffer from poor filtering power (which results in high verification cost), or incur too much computational cost to guarantee the filtering power. In this paper, we propose a multiple prefix filtering method based on different global orderings such that the number of candidate pairs can be reduced significantly. We also propose a parallel extension of the algorithm that is efficient and scalable in a MapReduce framework. We conduct extensive experiments on both centralized and Hadoop systems using both real and synthetic data sets, and the results show that our proposed approach outperforms existing approaches in both efficiency and scalability.

Graphs are popular models for representing complex structure data and similarity search for graphs has become a fundamental research problem. Many techniques have been proposed to support similarity search based on the graph edit distance. However, they all suffer from certain drawbacks: high computational complexity, poor scalability in terms of database size, or not taking full advantage of indexes. To address these problems, in this paper, we propose SEGOS, an indexing and query processing framework for graph similarity search. First, an effective two-level index is constructed off-line based on sub-unit decomposition of graphs. Then, a novel search strategy based on the index is proposed. Two algorithms adapted from TA and CA methods are seamlessly integrated into the proposed strategy to enhance graph search. More specially, the proposed framework is easy to be pipelined to support continuous graph pruning. Extensive experiments are conducted on two real datasets to evaluate the effectiveness and scalability of our approaches.