Computer Science Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/2756
Browse
3 results
Search Results
Item Evaluating Machine Intelligence with Question Answering(2021) rodriguez, pedro; Boyd-Graber, Jordan; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Humans ask questions to learn about the world and to test knowledge understanding. The ability to ask questions combines aspects of intelligence unique to humans: language understanding, knowledge representation, and reasoning. Thus, building systems capable of intelligent question answering (QA) is a grand goal of natural language processing (NLP). To measure progress in NLP, we create "exams" for computer systems and compare their effectiveness against a reference point---often based on humans. How precisely we measure progress depends on whether we are building computer systems that optimize human satisfaction in information-seeking tasks or that measure progress towards intelligent QA. In the first part of this dissertation, we explore each goal in turn, how they differ, and describe their relationship to QA formats. As an example of an information-seeking evaluation, we introduce a new dialog QA task paired with a new evaluation method. Afterward, we turn our attention to using QA to evaluate machine intelligence. A good evaluation should be able to discriminate between lesser and more capable QA models. This dissertation explores three ways to improve the discriminative power of QA evaluations: (1) dynamic weighting of test questions, (2) a format that by construction tests multiple levels of knowledge, and (3) evaluation data that is created through human-computer collaboration. By dynamically weighting test questions, we challenge a foundational assumption of the de facto standard in QA evaluation---the leaderboard. Namely, we contend that contrary to nearly all QA and NLP evaluations which implicitly assign equal weights to examples by averaging scores, that examples are not equally useful for estimating machine (or human) QA ability. As any student may tell you, not all questions on an exam are equally difficult and in the worst-case questions are unsolvable. Drawing on decades of research in educational testing, we propose adopting an alternative evaluation methodology---Item Response Theory---that is widely used to score human exams (e.g., the SAT). By dynamically weighting questions, we show that this improves the reliability of leaderboards in discriminating between models of differing QA ability while also being helpful in the construction of new evaluation datasets. Having improved the scoring of models, we next turn to improving the format and data in QA evaluations. Our idea is simple. In most QA tasks (e.g., Jeopardy!), each question tests a single level of knowledge; in our task (the trivia game Quizbowl), we test multiple levels of knowledge with each question. Since each question tests multiple levels of knowledge, this decreases the likelihood that we learn nothing about the difference between two models (i.e., they are both correct or both wrong), which substantially increases discriminative power. Despite the improved format, we next show that while our QA models defeat accomplished trivia players, that they are overly reliant on brittle pattern matching, which indicates a failure to intelligently answer questions. To mitigate this problem, we introduce a new framework for building evaluation data where humans and machines cooperatively craft trivia questions that are difficult to answer through clever pattern matching tricks alone---while being no harder for humans. We conclude by sketching a broader vision for QA evaluation that combines the three components of evaluation we improve---scoring, format, and data---to create living evaluations and re-imagine the role of leaderboards.Item Discourse-Level Language Understanding with Deep Learning(2017) Iyyer, Mohit Nagaraja; Boyd-Graber, Jordan; Daumé, Hal; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Designing computational models that can understand language at a human level is a foundational goal in the field of natural language processing (NLP). Given a sentence, machines are capable of translating it into many different languages, generating a corresponding syntactic parse tree, marking words that refer to people or places, and much more. These tasks are solved by statistical machine learning algorithms, which leverage patterns in large datasets to build predictive models. Many recent advances in NLP are due to deep learning models (parameterized as neural networks), which bypass user-specified features in favor of building representations of language directly from the text. Despite many deep learning-fueled advances at the word and sentence level, however, computers still struggle to understand high-level discourse structure in language, or the way in which authors combine and order different units of text (e.g., sentences, paragraphs, chapters) to express a coherent message or narrative. Part of the reason is data-related, as there are no existing datasets for many contextual language-based problems, and some tasks are too complex to be framed as supervised learning problems; for the latter type, we must either resort to unsupervised learning or devise training objectives that simulate the supervised setting. Another reason is architectural: neural networks designed for sentence-level tasks require additional functionality, interpretability, and efficiency to operate at the discourse level. In this thesis, I design deep learning architectures for three NLP tasks that require integrating information across high-level linguistic context: question answering, fictional relationship understanding, and comic book narrative modeling. While these tasks are very different from each other on the surface, I show that similar neural network modules can be used in each case to form contextual representations.Item A Neurocomputational Model of Grounded Language Comprehension and Production at the Sentence Level(2011) Monner, Derek; Reggia, James A; Computer Science; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)While symbolic and statistical approaches to natural language processing have become undeniably impressive in recent years, such systems still display a tendency to make errors that are inscrutable to human onlookers. This disconnect with human processing may stem from the vast differences in the substrates that underly natural language processing in artificial systems versus biological systems. To create a more relatable system, this dissertation turns to the more biologically inspired substrate of neural networks, describing the design and implementation of a model that learns to comprehend and produce language at the sentence level. The model's task is to ground simulated speech streams, representing a simple subset of English, in terms of a virtual environment. The model learns to understand and answer full-sentence questions about the environment by mimicking the speech stream of another speaker, much as a human language learner would. It is the only known neural model to date that can learn to map natural language questions to full-sentence natural language answers, where both question and answer are represented sublexically as phoneme sequences. The model addresses important points for which most other models, neural and otherwise, fail to account. First, the model learns to ground its linguistic knowledge using human-like sensory representations, gaining language understanding at a deeper level than that of syntactic structure. Second, analysis provides evidence that the model learns combinatorial internal representations, thus gaining the compositionality of symbolic approaches to cognition, which is vital for computationally efficient encoding and decoding of meaning. The model does this while retaining the fully distributed representations characteristic of neural networks, providing the resistance to damage and graceful degradation that are generally lacking in symbolic and statistical approaches. Finally, the model learns via direct imitation of another speaker, allowing it to emulate human processing with greater fidelity, thus increasing the relatability of its behavior. Along the way, this dissertation develops a novel training algorithm that, for the first time, requires only local computations to train arbitrary second-order recurrent neural networks. This algorithm is evaluated on its overall efficacy, biological feasibility, and ability to reproduce peculiarities of human learning such as age-correlated effects in second language acquisition.