Text-based Models Research Articles

Long-term stock index forecasting based on text mining of regulatory disclosures

Share valuations are known to adjust to new information entering the market, such as regulatory disclosures. We study whether the language of such news items can improve short-term and especially long-term (24 months) forecasts of stock indices. For this purpose, this work utilizes predictive models suited to high-dimensional data and specifically compares techniques for data-driven and knowledge-driven dimensionality reduction in order to avoid overfitting. Our experiments, based on 75,927 ad hoc announcements from 1996–2016, reveal the following results: in the long run, text-based models succeed in reducing forecast errors below baseline predictions from historic lags at a statistically significant level. Our research provides implications to business applications of decision-support in financial markets, especially given the growing prevalence of index ETFs (exchange traded funds).

Product Life Cycles in Corporate Finance

We develop a novel 10-K text-based model of product life cycles and examine firm investment policies. Conditioning on the life cycle substantially improves the power of Q to explain investment, and reveals a natural ordering of investments over the life cycle. While R&D and CAPX sensitivity are high early in the cycle, acquisitions arise as firms mature, and divestitures and product extension investments arise as firms decline. Q sensitivities that condition on the life cycle can vary by as much as 400% from traditional sensitivities. The life cycle framework further reveals an enriched relationship between competition, investment and corporate profits.

Learning Multimodal Word Representation via Dynamic Fusion Methods

Multimodal models have been proven to outperform text-based models on learning semantic word representations. Almost all previous multimodal models typically treat the representations from different modalities equally. However, it is obvious that information from different modalities contributes differently to the meaning of words. This motivates us to build a multimodal model that can dynamically fuse the semantic representations from different modalities according to different types of words. To that end, we propose three novel dynamic fusion methods to assign importance weights to each modality, in which weights are learned under the weak supervision of word association pairs. The extensive experiments have demonstrated that the proposed methods outperform strong unimodal baselines and state-of-the-art multimodal models.

Visually Grounded and Textual Semantic Models Differentially Decode Brain Activity Associated with Concrete and Abstract Nouns

Important advances have recently been made using computational semantic models to decode brain activity patterns associated with concepts; however, this work has almost exclusively focused on concrete nouns. How well these models extend to decoding abstract nouns is largely unknown. We address this question by applying state-of-the-art computational models to decode functional Magnetic Resonance Imaging (fMRI) activity patterns, elicited by participants reading and imagining a diverse set of both concrete and abstract nouns. One of the models we use is linguistic, exploiting the recent word2vec skipgram approach trained on Wikipedia. The second is visually grounded, using deep convolutional neural networks trained on Google Images. Dual coding theory considers concrete concepts to be encoded in the brain both linguistically and visually, and abstract concepts only linguistically. Splitting the fMRI data according to human concreteness ratings, we indeed observe that both models significantly decode the most concrete nouns; however, accuracy is significantly greater using the text-based models for the most abstract nouns. More generally this confirms that current computational models are sufficiently advanced to assist in investigating the representational structure of abstract concepts in the brain.

Opinion mining using ensemble text hidden Markov models for text classification

With the rapid growth of social media, text mining is extensively utilized in practical fields, and opinion mining, also known as sentiment analysis, plays an important role in analyzing opinion and sentiment in texts. Methods in opinion mining generally depend on a sentiment lexicon, which is a set of predefined key words that express sentiment. Opinion mining requires proper sentiment words to be extracted in advance and has difficulty classifying sentences that imply an opinion without using any sentiment key words. This paper presents a new sentiment analysis method, based on text-based hidden Markov models (TextHMMs), for text classification that uses a sequence of words in training texts instead of a predefined sentiment lexicon. We sought to learn text patterns representing sentiment through ensemble TextHMMs. Our method defines hidden variables in TextHMMs by semantic cluster information in consideration of the co-occurrence of words, and thus calculates the sentiment orientation of sentences by fitted TextHMMs. To reflect diverse patterns, we applied an ensemble of TextHMM-based classifiers. In the experiments with a benchmark data set, we show that this method is superior to some existing methods and particularly has potential to classify implicit opinions. We also demonstrate the practicality of the proposed method in a real-life data set of online market reviews.

QUERY SUPPORT FOR GMZ

Abstract. Generic text-based compression models are simple and fast but there are two issues that needs to be addressed. They cannot leverage the structure that exists in data to achieve better compression and there is an unnecessary decompression step before the user can actually use the data. To address these issues, we came up with GMZ, a lossless compression model aimed at achieving high compression ratios. The decision to design GMZ (Khandelwal and Rajan, 2017) exclusively for GML's Simple Features Profile (SFP) seems fair because of the high use of SFP in WFS and that it facilitates high optimisation of the compression model. This is an extension of our work on GMZ. In a typical server-client model such as Web Feature Service, the server is the primary creator and provider of GML, and therefore, requires compression and query capabilities. On the other hand, the client is the primary consumer of GML, and therefore, requires decompression and visualisation capabilities. In the first part of our work, we demonstrated compression using a python script that can be plugged in a server architecture, and decompression and visualisation in a web browser using a Firefox addon. The focus of this work is to develop the already existing tools to provide query capability to server. Our model provides the ability to decompress individual features in isolation, which is an essential requirement for realising query in compressed state. We con - struct an R-Tree index for spatial data and a custom index for non-spatial data and store these in a separate index file to prevent alter - ing the compression model. This facilitates independent use of compressed GMZ file where index can be constructed when required. The focus of this work is the bounding-box or range query commonly used in webGIS with provision for other spatial and non-spatial queries. The decrement in compression ratios due to the new index file is in the range of 1–3 percent which is trivial considering the benefits of querying in compressed state. With around 75 % average compression of the original data, query support in compressed state and decompression support in the browser, GMZ can be a good alternative to GML for WFS-like services.

An IFC schema extension and binary serialization format to efficiently integrate point cloud data into building models

In this paper we suggest an extension to the Industry Foundation Classes (IFC) model to integrate point cloud datasets. The proposal includes a schema extension to the core model allowing the storage of points, either as Cartesian coordinates, points in parametric space of associated building element surfaces or as discrete height fields projected as grids onto building elements. To handle the considerable amounts of data generated in the process of scanning building structures, we present intelligent compression approaches combined with the Hierarchical Data Format (HDF) as an efficient serialization and an alternative to clear text encoded ISO 10303 part 21 files. Based on prototypical implementations we show results of various serialization options and their impacts on storage efficiency.In this proposal the deepened semantic relationships have been favoured over compression ratios. Nevertheless, with various near-lossless layers of compression and binary serialization applied, a compression ratio of up to 67.7% is obtained for a building model with integrated point clouds, compared to the raw source data. The binary serialization is able to handle hundreds of millions of points, out of which specific spatial and semantic subsets can rapidly be extracted due to the containerized hierarchical storage model and association to building elements. The authors advocate the use of binary storage for sizeable point cloud scans, but also show how especially the grid discretization can result into usable points cloud segments embedded into text-based IFC models.

Clustering documents in evolving languages by image texture analysis

This paper introduces a new method for clustering of documents, which have been written in a language evolving during different historical periods, with an example of the Italian language. In the first phase, the text is transformed into a string of four numerical codes, which have been derived from the energy profile of each letter, defining the height of the letters and their location in the text line. Each code represents a gray level and the text is codified as a 1-D image. In the second phase, texture features are extracted from the obtained image in order to create document feature vectors. Subsequently, a new clustering algorithm is employed on the feature vectors to discriminate documents from different historical periods of the language. Experiments are performed on a database of Italian documents given in Italian Vulgar and modern Italian. Results demonstrate that this proposed method perfectly identifies the historical periods of the language of the documents, outperforming other well-known clustering algorithms generally adopted for document categorization and other state-of-the-art text-based language models.

Visually Grounded Meaning Representations.

In this paper we address the problem of grounding distributional representations of lexical meaning. We introduce a new model which uses stacked autoencoders to learn higher-level representations from textual and visual input. The visual modality is encoded via vectors of attributes obtained automatically from images. We create a new large-scale taxonomy of 600 visual attributes representing more than 500 concepts and 700 K images. We use this dataset to train attribute classifiers and integrate their predictions with text-based distributional models of word meaning. We evaluate our model on its ability to simulate word similarity judgments and concept categorization. On both tasks, our model yields a better fit to behavioral data compared to baselines and related models which either rely on a single modality or do not make use of attribute-based input.

The cognitive plausibility of statistical classification models: Comparing textual and behavioral evidence

AbstractUsage-based linguistics abounds with studies that use statistical classification models to analyze either textual corpus data or behavioral experimental data. Yet, before we can draw conclusions from statistical models of empirical data that we can feed back into cognitive linguistic theory, we need to assess whether the text-based models are cognitively plausible and whether the behavior-based models are linguistically accurate. In this paper, we review four case studies that evaluate statistical classification models of richly annotated linguistic data by explicitly comparing the performance of a corpus-based model to the behavior of native speakers. The data come from four different languages (Arabic, English, Estonian, and Russian) and pertain to both lexical as well as syntactic near-synonymy. We show that behavioral evidence is needed in order to fine tune and improve statistical models built on data from a corpus. We argue that methodological pluralism is the key for a cognitively realistic linguistic theory.

"Rate My Therapist": Automated Detection of Empathy in Drug and Alcohol Counseling via Speech and Language Processing.

The technology for evaluating patient-provider interactions in psychotherapy–observational coding–has not changed in 70 years. It is labor-intensive, error prone, and expensive, limiting its use in evaluating psychotherapy in the real world. Engineering solutions from speech and language processing provide new methods for the automatic evaluation of provider ratings from session recordings. The primary data are 200 Motivational Interviewing (MI) sessions from a study on MI training methods with observer ratings of counselor empathy. Automatic Speech Recognition (ASR) was used to transcribe sessions, and the resulting words were used in a text-based predictive model of empathy. Two supporting datasets trained the speech processing tasks including ASR (1200 transcripts from heterogeneous psychotherapy sessions and 153 transcripts and session recordings from 5 MI clinical trials). The accuracy of computationally-derived empathy ratings were evaluated against human ratings for each provider. Computationally-derived empathy scores and classifications (high vs. low) were highly accurate against human-based codes and classifications, with a correlation of 0.65 and F-score (a weighted average of sensitivity and specificity) of 0.86, respectively. Empathy prediction using human transcription as input (as opposed to ASR) resulted in a slight increase in prediction accuracies, suggesting that the fully automatic system with ASR is relatively robust. Using speech and language processing methods, it is possible to generate accurate predictions of provider performance in psychotherapy from audio recordings alone. This technology can support large-scale evaluation of psychotherapy for dissemination and process studies.

CrossLink

Collections of images and 3D models hide in them many interesting aspects of our surroundings. Significant efforts have been devoted to organize and explore such data repositories. Most such efforts, however, process the two data modalities separately, and do not take full advantage of the complementary information that exist in different domains, which can help to solve difficult problems in one by exploiting the structure in the other. Beyond the obvious difference in data representations, a key difficulty in such joint analysis lies in the significant variability in the structure and inherent properties of the 2D and 3D data collections, which hinders cross-domain analysis and exploration. We introduce CrossLink, a system for joint image-3D model processing that uses the complementary strengths of each data modality to facilitate analysis and exploration. We first show how our system significantly improves the quality of text-based 3D model search by using side information coming from an image database. We then demonstrate how to consistently align the filtered 3D model collections, and then use them to re-sort image collections based on pose and shape attributes. We evaluate our framework both quantitatively and qualitatively on 20 object categories of 2D image and 3D model collections, and quantitatively demonstrate how a wide variety of tasks in each data modality can strongly benefit from the complementary information present in the other, paving the way to a richer 2D and 3D processing toolbox.

Reading visually embodied meaning from the brain: Visually grounded computational models decode visual-object mental imagery induced by written text

Embodiment theory predicts that mental imagery of object words recruits neural circuits involved in object perception. The degree of visual imagery present in routine thought and how it is encoded in the brain is largely unknown. We test whether fMRI activity patterns elicited by participants reading objects' names include embodied visual-object representations, and whether we can decode the representations using novel computational image-based semantic models. We first apply the image models in conjunction with text-based semantic models to test predictions of visual-specificity of semantic representations in different brain regions. Representational similarity analysis confirms that fMRI structure within ventral-temporal and lateral-occipital regions correlates most strongly with the image models and conversely text models correlate better with posterior-parietal/lateral-temporal/inferior-frontal regions. We use an unsupervised decoding algorithm that exploits commonalities in representational similarity structure found within both image model and brain data sets to classify embodied visual representations with high accuracy (8/10) and then extend it to exploit model combinations to robustly decode different brain regions in parallel. By capturing latent visual-semantic structure our models provide a route into analyzing neural representations derived from past perceptual experience rather than stimulus-driven brain activity. Our results also verify the benefit of combining multimodal data to model human-like semantic representations.

Grounding Action Descriptions in Videos

Recent work has shown that the integration of visual information into text-based models can substantially improve model predictions, but so far only visual information extracted from static images has been used. In this paper, we consider the problem of grounding sentences describing actions in visual information extracted from videos. We present a general purpose corpus that aligns high quality videos with multiple natural language descriptions of the actions portrayed in the videos, together with an annotation of how similar the action descriptions are to each other. Experimental results demonstrate that a text-based model of similarity between actions improves substantially when combined with visual information from videos depicting the described actions.

Unsupervised Dependency Parsing with Acoustic Cues

Unsupervised parsing is a difficult task that infants readily perform. Progress has been made on this task using text-based models, but few computational approaches have considered how infants might benefit from acoustic cues. This paper explores the hypothesis that word duration can help with learning syntax. We describe how duration information can be incorporated into an unsupervised Bayesian dependency parser whose only other source of information is the words themselves (without punctuation or parts of speech). Our results, evaluated on both adult-directed and child-directed utterances, show that using word duration can improve parse quality relative to words-only baselines. These results support the idea that acoustic cues provide useful evidence about syntactic structure for language-learning infants, and motivate the use of word duration cues in NLP tasks with speech.

“Voices of the People”: Linguistic Research Among Germany's Prisoners of War During World War I

This paper investigates the history of the Royal Prussian Phonographic Commission, a body that collected and archived linguistic, ethnographic, and anthropological data from prisoners-of-war (POWs) in Germany during World War I. Recent literature has analyzed the significance of this research for the rise of conservative physical anthropology. Taking a complementary approach, the essay charts new territory in seeking to understand how the prison-camp studies informed philology and linguistics specifically. I argue that recognizing philological commitments of the Phonographic Commission is essential to comprehending the project contextually. My approach reveals that linguists accommodated material and contemporary evidence to older text-based research models, sustaining dynamic theories of language. Through a case study based on the Iranian philologist F. C. Andreas (1846-1930), the paper ultimately argues that linguistics merits greater recognition in the historiography of the behavioral sciences.

Securing the Past: Conservation in Art, Architecture and Literature

Securing the Past: Conservation in Art, Architecture and Literature

Securing the Past: Conservation in Art, Architecture and Literature

Securing the Past: Conservation in Art, Architecture and Literature

OiNC: A Comprehensive CAD Import and Tracking System for Monte Carlo Radiation Transport Calculations

A computer-aided design (CAD) import and tracking system, OiNC, has been developed that is currently being incorporated into the MONK and MCBEND criticality and shielding codes. The system enables analysts to take advantage of the advanced geometry modeling capabilities provided by the solid CAD modelers Autodesk Inventor and Solidworks. Various tracking acceleration techniques are implemented in order to reduce calculation time, including the full or partial automatic conversion of the CAD model to constructive solid geometry (CSG) form. For the criticality benchmark model detailed here, the overall calculation run time for a CAD-based geometry was found to be just 14% longer than an equivalent text-based CSG model. OiNC uses analytical surface descriptions, with full NURBS support, avoiding the approximations and problems associated with meshing. However, meshed models can still be imported and tracked, enabling shielding and criticality calculations to be performed, for example, on impact simulated finite element models.

MQSS: multimodal query suggestion and searching for video search

In this paper, we propose a multimodal query suggestion method for video search which can leverage multimodal processing to improve the quality of search results. When users type general or ambiguous textual queries, our system MQSS provides keyword suggestions and representative image examples in an easy-to-use dropdown manner which can help users specify their search intent more precisely and effortlessly. It is a powerful complement to initial queries. After the queries are formulated as multimodal query (i.e., text, image), the new queries are input to individual search models, such as text-based, concept-based and visual example-based search model. Then we apply multimodal fusion method to aggregate the above-mentioned several search results. The effectiveness of MQSS is demonstrated by evaluations over a web video data set.