Human-annotated rationales and explainable text classification: a survey.

Asking annotators to explain "why" they labeled an instance yields annotator rationales: natural language explanations that provide reasons for classifications. In this work, we survey the collection and use of annotator rationales. Human-annotated rationales can improve data quality and form a valuable resource for improving machine learning models. Moreover, human-annotated rationales can inspire the construction and evaluation of model-annotated rationales, which can play an important role in explainable artificial intelligence.

Interpreting vision and language generative models with semantic visual priors.

When applied to Image-to-text models, explainability methods have two challenges. First, they often provide token-by-token explanations namely, they compute a visual explanation for each token of the generated sequence. This makes explanations expensive to compute and unable to comprehensively explain the model's output. Second, for models with visual inputs, explainability methods such as SHAP typically consider superpixels as features. Since superpixels do not correspond to semantically meaningful regions of an image, this makes explanations harder to interpret. We develop a framework based on SHAP, that allows for generating comprehensive, meaningful explanations leveraging the meaning representation of the output sequence as a whole. Moreover, by exploiting semantic priors in the visual backbone, we extract an arbitrary number of features that allows the efficient computation of Shapley values on large-scale models, generating at the same time highly meaningful visual explanations. We demonstrate that our method generates semantically more expressive explanations than traditional methods at a lower compute cost and that it can be generalized to a large family of vision-language models.

Conceptualization in reference production: Probabilistic modeling and experimental testing.

In psycholinguistics, there has been relatively little work investigating conceptualization-how speakers decide which concepts to express. This contrasts with work in natural language generation (NLG), a subfield of artificial intelligence, where much research has explored content determination during the generation of referring expressions. Existing NLG algorithms for conceptualization during reference production do not fully explain previous psycholinguistic results, so we developed new models that we tested in three language production experiments. In our experiments, participants described target objects to another participant. In Experiment 1, either size, color, or both distinguished the target from all distractor objects; in Experiment 2, either color, type, or both color and type distinguished it from all distractors; In Experiment 3, color, size, or the border around the object distinguished the target. We tested how well the different models fit the distribution of description types (e.g., "small candle," "gray candle," "small gray candle") that participants produced. Across these experiments, the probabilistic referential overspecification model (PRO) provided the best fit. In this model, speakers first choose a property that rules out all distractors. If there is more than one such property, then they probabilistically choose one on the basis of a preference for that property. Next, they sometimes add another property, with the probability again determined by its preference and speakers' eagerness to overspecify. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Reference Production as Search: The Impact of Domain Size on the Production of Distinguishing Descriptions.

When producing a description of a target referent in a visual context, speakers need to choose a set of properties that distinguish it from its distractors. Computational models of language production/generation usually model this as a search process and predict that the time taken will increase both with the number of distractors in a scene and with the number of properties required to distinguish the target. These predictions are reminiscent of classic findings in visual search; however, unlike models of reference production, visual search models also predict that search can become very efficient under certain conditions, something that reference production models do not consider. This paper investigates the predictions of these models empirically. In two experiments, we show that the time taken to plan a referring expression-as reflected by speech onset latencies-is influenced by distractor set size and by the number of properties required, but this crucially depends on the discriminability of the properties under consideration. We discuss the implications for current models of reference production and recent work on the role of salience in visual search.

Production of Referring Expressions for an Unknown Audience: A Computational Model of Communal Common Ground.

This article presents a computational model of the production of referring expressions under uncertainty over the hearer's knowledge. Although situations where the hearer's knowledge is uncertain have seldom been addressed in the computational literature, they are common in ordinary communication, for example when a writer addresses an unknown audience, or when a speaker addresses a stranger. We propose a computational model composed of three complimentary heuristics based on, respectively, an estimation of the recipient's knowledge, an estimation of the extent to which a property is unexpected, and the question of what is the optimum number of properties in a given situation. The model was tested in an experiment with human readers, in which it was compared against the Incremental Algorithm and human-produced descriptions. The results suggest that the new model outperforms the Incremental Algorithm in terms of the proportion of correctly identified entities and in terms of the perceived quality of the generated descriptions.

Managing ambiguity in reference generation: the role of surface structure.

This article explores the role of surface ambiguities in referring expressions, and how the risk of such ambiguities should be taken into account by an algorithm that generates referring expressions, if these expressions are to be optimally effective for a hearer. We focus on the ambiguities that arise when adjectives occur in coordinated structures. The central idea is to use statistical information about lexical co-occurrence to estimate which interpretation of a phrase is most likely for human readers, and to avoid generating phrases where misunderstandings are likely. Various aspects of the problem were explored in three experiments in which responses by human participants provided evidence about which reading was most likely for certain phrases, which phrases were deemed most suitable for particular referents, and the speed at which various phrases were read. We found a preference for ''clear'' expressions to ''unclear'' ones, but if several of the expressions are ''clear,'' then brief expressions are preferred over non-brief ones even though the brief ones are syntactically ambiguous and the non-brief ones are not; the notion of clarity was made precise using Kilgarriff's Word Sketches. We outline an implemented algorithm that generates noun phrases conforming to our hypotheses.

Toward a computational psycholinguistics of reference production.

This article introduces the topic ''Production of Referring Expressions: Bridging the Gap between Computational and Empirical Approaches to Reference'' of the journal Topics in Cognitive Science. We argue that computational and psycholinguistic approaches to reference production can benefit from closer interaction, and that this is likely to result in the construction of algorithms that differ markedly from the ones currently known in the computational literature. We focus particularly on determinism, the feature of existing algorithms that is perhaps most clearly at odds with psycholinguistic results, discussing how future algorithms might include non-determinism, and how new psycholinguistic experiments could inform the development of such algorithms.

Generation of referring expressions: assessing the Incremental Algorithm.

A substantial amount of recent work in natural language generation has focused on the generation of ''one-shot'' referring expressions whose only aim is to identify a target referent. Dale and Reiter's Incremental Algorithm (IA) is often thought to be the best algorithm for maximizing the similarity to referring expressions produced by people. We test this hypothesis by eliciting referring expressions from human subjects and computing the similarity between the expressions elicited and the ones generated by algorithms. It turns out that the success of the IA depends substantially on the ''preference order'' (PO) employed by the IA, particularly in complex domains. While some POs cause the IA to produce referring expressions that are very similar to expressions produced by human subjects, others cause the IA to perform worse than its main competitors; moreover, it turns out to be difficult to predict the success of a PO on the basis of existing psycholinguistic findings or frequencies in corpora. We also examine the computational complexity of the algorithms in question and argue that there are no compelling reasons for preferring the IA over some of its main competitors on these grounds. We conclude that future research on the generation of referring expressions should explore alternatives to the IA, focusing on algorithms, inspired by the Greedy Algorithm, which do not work with a fixed PO.