The goal of interpretable machine learning (ML) is to design tools and visualizations to help users scrutinize a system’s predictions. Prior studies have mostly employed quantitative methods to investigate the effects of specific tools/visualizations on outcomes related to objective performance—a human’s ability to correctly agree or disagree with the system—and subjective perceptions of the system. Few studies have employed qualitative methods to investigate how and why specific tools/visualizations influence performance, perceptions, and behaviors. We report on a lab study (N = 30) that investigated the influences of two interpretability features: confidence values and sentence highlighting. Participants judged whether medical articles belong to a predicted medical topic and were exposed to two interface conditions—one with and one without interpretability features. We investigate the effects of our interpretability features on participants’ performance and perceptions. Additionally, we report on a qualitative analysis of participants’ responses during an exit interview. Specifically, we report on how our interpretability features impacted different cognitive activities that participants engaged with during the task—reading, learning, and decision making. We also describe ways in which the interpretability features introduced challenges and sometimes led participants to make mistakes. Insights gained from our results point to future directions for interpretable ML research.

The United States Environmental Protection Agency (EPA) periodically releases Integrated Science Assessments (ISAs) that synthesize the latest research on each of six air pollutants to inform environmental policymaking. To guarantee the best possible coverage of relevant literature, EPA scientists spend months manually screening hundreds of thousands of references to identify a small proportion to be cited in an ISA. The challenge of extreme scale and the pursuit of maximum recall calls for effective machine-assisted approaches to reducing the time and effort required by the screening process. This work introduces the ISA literature screening dataset and the associated research challenges to the information and knowledge management community. Our pilot experiments show that combining multiple approaches in tackling this challenge is both promising and necessary. The dataset is available at https://catalog.data.gov/dataset/isa-literature-screening-dataset-v-1.

Faceted search systems enable users to filter results by selecting values along different dimensions or facets. Traditionally, facets have corresponded to properties of information items that are part of the document metadata. Recently, faceted search systems have begun to use machine learning to automatically associate documents with facet-values that are more subjective and abstract. Examples include search systems that support topic-based filtering of research articles, concept-based filtering of medical documents, and tag-based filtering of images. While machine learning can be used to infer facet-values when the collection is too large for manual annotation, machine-learned classifiers make mistakes. In such cases, it is desirable to have a scrutable system that explains why a filtered result is relevant to a facet-value. Such explanations are missing from current systems. In this paper, we investigate how explainability features can help users interpret results filtered using machine-learned facets. We consider two explainability features: (1) showing prediction confidence values and (2) highlighting rationale sentences that played an influential role in predicting a facet-value. We report on a crowdsourced study involving 200 participants. Participants were asked to scrutinize movie plot summaries predicted to satisfy multiple genres and indicate their agreement or disagreement with the system. Participants were exposed to four interface conditions. We found that both explainability features had a positive impact on participants' perceptions and performance. While both features helped, the sentence-highlighting feature played a more instrumental role in enabling participants to reject false positive cases. We discuss implications for designing tools to help users scrutinize automatically assigned facet-values.

We study the problem of incorporating prior knowledge into a deep Transformer-based model, i.e., Bidirectional Encoder Representations from Transformers (BERT), to enhance its performance on semantic textual matching tasks. By probing and analyzing what BERT has already known when solving this task, we obtain better understanding of what task-specific knowledge BERT needs the most and where it is most needed. The analysis further motivates us to take a different approach than most existing works. Instead of using prior knowledge to create a new training task for fine-tuning BERT, we directly inject knowledge into BERT’s multi-head attention mechanism. This leads us to a simple yet effective approach that enjoys fast training stage as it saves the model from training on additional data or tasks other than the main task. Extensive experiments demonstrate that the proposed knowledge-enhanced BERT is able to consistently improve semantic textual matching performance over the original BERT model, and the performance benefit is most salient when training data is scarce.

In professional search tasks such as precision medicine literature search, queries often involve multiple aspects. To assess the relevance of a document, a searcher often painstakingly validates each aspect in the query and follows a task-specific logic to make a relevance decision. In such scenarios, we say the searcher makes a structured relevance judgment, as opposed to the traditional univariate (binary or graded) relevance judgment. Ideally, a search engine can support searcher's workflow and follow the same steps to predict document relevance. This approach may not only yield highly effective retrieval models, but also open up opportunities for the model to explain its decision in the same "lingo" as the searcher. Using structured relevance judgment data from the TREC Precision Medicine track, we propose novel retrieval models that emulate how medical experts make structured relevance judgments. Our experiments demonstrate that these simple, explainable models can outperform complex, black-box learning-to-rank models.

Modern search engines aggregate results from different verticals : webpages, news, images, video, shopping, knowledge cards, local maps, and so on. Unlike “ten blue links,” these search results are heterogeneous in nature and not even arranged in a list on the page. This revolution directly challenges the conventional “ranked list” formulation in ad hoc search. Therefore, finding proper presentation for a gallery of heterogeneous results is critical for modern search engines. We propose a novel framework that learns the optimal page presentation to render heterogeneous results onto search result page (SERP). Page presentation is broadly defined as the strategy to present a set of items on SERP, much more expressive than a ranked list. It can specify item positions, image sizes, text fonts, and any other styles as long as variations are within business and design constraints. The learned presentation is content aware, i.e., tailored to specific queries and returned results. Simulation experiments show that the framework automatically learns eye-catchy presentations for relevant results. Experiments on real data show that simple instantiations of the framework already outperform leading algorithm in federated search result presentation. It means the framework can learn its own result presentation strategy purely from data, without even knowing the “probability ranking principle.”

Effectively making sense of short texts is a critical task for many real world applications such as search engines, social media services, and recommender systems. The task is particularly challenging as a short text contains very sparse information, often too sparse for a machine learning algorithm to pick up useful signals. A common practice for analyzing short text is to first expand it with external information, which is usually harvested from a large collection of longer texts. In literature, short text expansion has been done with all kinds of heuristics. We propose an end-to-end solution that automatically learns how to expand short text to optimize a given learning task. A novel deep memory network is proposed to automatically find relevant information from a collection of longer documents and reformulate the short text through a gating mechanism. Using short text classification as a demonstrating task, we show that the deep memory network significantly outperforms classical text expansion methods with comprehensive experiments on real world data sets.

Modern search engines aggregate results from different verticals: webpages, news, images, video, shopping, knowledge cards, local maps, etc. Unlike "ten blue links", these search results are heterogeneous in nature and not even arranged in a list on the page. This revolution directly challenges the conventional "ranked list" formulation in ad hoc search. Therefore, finding proper presentation for a gallery of heterogeneous results is critical for modern search engines.

We consider a scenario where a searcher requires both high precision and high recall from an interactive retrieval process. Such scenarios are very common in real life, exemplified by medical search, legal search, market research, and literature review. When access to the entire data set is available, an active learning loop could be used to ask for additional relevance feedback labels in order to refine a classifier. When data is accessed via search services, however, only limited subsets of the corpus can be considered, subsets defined by queries. In that setting, relevance feedback has been used in a query enhancement loop that updates a query.