Large Inventory Research Articles

LMMS reloaded: Transformer-based sense embeddings for disambiguation and beyond

Distributional semantics based on neural approaches is a cornerstone of Natural Language Processing, with surprising connections to human meaning representation as well. Recent Transformer-based Language Models have proven capable of producing contextual word representations that reliably convey sense-specific information, simply as a product of self-supervision. Prior work has shown that these contextual representations can be used to accurately represent large sense inventories as sense embeddings, to the extent that a distance-based solution to Word Sense Disambiguation (WSD) tasks outperforms models trained specifically for the task. Still, there remains much to understand on how to use these Neural Language Models (NLMs) to produce sense embeddings that can better harness each NLM's meaning representation abilities. In this work we introduce a more principled approach to leverage information from all layers of NLMs, informed by a probing analysis on 14 NLM variants. We also emphasize the versatility of these sense embeddings in contrast to task-specific models, applying them on several sense-related tasks, besides WSD, while demonstrating improved performance using our proposed approach over prior work focused on sense embeddings. Finally, we discuss unexpected findings regarding layer and model performance variations, and potential applications for downstream tasks.

Balancing earliness and tardiness within workload control order release: an assessment by simulation

Most Workload Control literature assumes that delivery performance is determined by tardiness related performance measures only. While this may be true for companies that directly deliver to end-customers, for make-to-stock companies or firms that are part of supply chains, producing early often means large inventories in the finished goods warehouse or penalties incurred by companies downstream in the supply chain. Some earlier Workload Control studies used a so-called time limit, which constrains the set of jobs that can be considered for order release, to reduce earliness. However, recent literature largely abandoned the time limit since it negatively impacts tardiness performance. This study revisits the time limit, assessing the use of different adaptive policies that restrict its use to periods of either low or high load. By using a simulation model of a pure job shop, the study shows that an adaptive policy allows to balance the contradictory objectives of delaying the release of orders to reduce earliness and to release orders early to respond to periods of high load as quick as possible. Meanwhile, only using a time limit in periods of high load was found to be the best policy.

Estimation of Temporal Variation of Discharged Tritium from Port of Fukushima Dai-ichi Nuclear Power Plant: Analysis of the Temporal Variation and Comparison with released Tritium from Japan and Major Nuclear Facilities Worldwide

We estimate the monthly discharge inventory of tritium from the port of Fukushima Daiichi Nuclear Power Plant (1F) from Jun. 2013 to Mar. 2020 using the Voronoi tessellation scheme, following the tritium monitoring inside the port that started in Jun. 2013. As for the missing period from the initial month, Apr. 2011 to May 2013, we calculate the tritium discharge by utilizing the ratio of tritium concentration to 137Cs concentration in stagnant contaminant water during the initial direct run-off period to Jun. 2011 and the discharge inventory correlation between tritium and 137Cs for the next-unknown continuous-discharge period up to May 2013. From all the estimated results over 9 years, we found that the monthly discharge inventory sharply dropped immediately after closing the seaside impermeable wall in Oct. 2015 and subsequently coincided well with the sum of those of drainage and subdrain etc. By comparing the estimated results with those in the normal operation period before the accident, we point out that the discharge inventory from the 1F port after the accident is not very large. Even the estimation for the year 2011 is found to be comparable to the maximum of operating pressurized water reactors releasing relatively large inventories in the number of digits. In the national level, the total domestic release inventory in Japan significantly decreased after the accident owing to the operational shutdown of most plants. Furthermore, 1F and even the total Japanese discharge inventory are found to be minor compared with those of nuclear reprocessing plants and heavy-water reactors on a worldwide level. From the above, we suggest that various scenarios can be openly discussed regarding the management of tritium stored inside 1F with the help of the present estimated data and its comparison with the past discharge inventory.

Dynamic Pricing for a Large Inventory of Substitutable Goods

Dynamic Pricing for a Large Inventory of Substitutable Goods

The inventory of Michael Meyer’s property (1758) as a reflection of a Tallinn (Reval) merchant’s material world during the Age of the Enlightenment

The inventory of Tallinn merchant Michael Meyer’s (1704–1758) property is one of the largest inventories of an 18th century citizen of Tallinn. Almost the entire world of his possessions is reflected in this unique source. The inventory provides a comprehensive picture of his success, lifestyle, and hobbies, and the diverse list of household items provides a good idea of a prosperous merchant’s home in northeast Europe in the 18th century. The unique body of sources (Michael Meyer’s will, property inventory, and auction reports) provides comprehensive insight into the development of Tallinn’s material culture, i.e., the material culture history of Northern Europe, during the century of Enlightenment.

Mapping tropical forest trees across large areas with lightweight cost-effective terrestrial laser scanning

Key messageWe used lightweight terrestrial laser scanning (TLS) to detect over 3000 stems per hectare across a 12-ha permanent forest plot in French Guiana, 81% of them < 10 cm in trunk diameter. This method retrieved 85% of the trees of a classic inventory. Finally, TLS revealed that stem positions of the classic inventory had geolocation errors of up to 6 m.ContextAccurate position mapping of tropical rainforest trees is crucial for baseline studies of tropical forest ecology but is labor-intensive. Terrestrial lidar scanning (TLS) is broadly used in temperate forest inventories, but its use in rainforests is restricted to the determination of individual tree volumes within small survey areas.AimsMapping tree stems across one large (12-ha) rainforest plot, including trees less than 10 cm DBH, and evaluating the precision of traditional mapping approaches.MethodsWe used lightweight TLS, co-registered the acquisitions, and developed a new efficient algorithm to process the TLS data.ResultsWe detected 36,422 stems of which 29,665 (81%) were < 10 cm in diameter at breast height (DBH). Of the trees ≥ 10 cm DBH previously censused in the plot, 85% were identified by TLS. Automatic DBH estimation from TLS data had an RMSE of 6 cm. RMSE was improved to 3 cm by a manual verification of the shape and quality of the stem points. The initial census map had substantial bias in tree geolocation with a maximum value around 6 m.ConclusionLightweight TLS technology is a promising tool for the estimation of stem tapering and volume. Here, we show that it also facilitates the establishment of large tropical forest inventories, by improving the positioning of trees, thus increasing the accuracy of forest inventories and their cost-effectiveness.

Starburst Energy Feedback Seen through HCO+/HOC+ Emission in NGC 253 from ALCHEMI

Abstract Molecular abundances are sensitive to the UV photon flux and cosmic-ray ionization rate. In starburst environments, the effects of high-energy photons and particles are expected to be stronger. We examine these astrochemical signatures through multiple transitions of HCO+ and its metastable isomer HOC+ in the center of the starburst galaxy NGC 253 using data from the Atacama Large Millimeter/submillimeter Array large program ALMA Comprehensive High-resolution Extragalactic Molecular inventory. The distribution of the HOC+(1−0) integrated intensity shows its association with “superbubbles,” cavities created either by supernovae or expanding H ii regions. The observed HCO+/HOC+ abundance ratios are ∼10–150, and the fractional abundance of HOC+ relative to H2 is ∼1.5 × 10−11–6 × 10−10, which implies that the HOC+ abundance in the center of NGC 253 is significantly higher than in quiescent spiral arm dark clouds in the Galaxy and the Galactic center clouds. Comparison with chemical models implies either an interstellar radiation field of G 0 ≳ 103 if the maximum visual extinction is ≳5, or a cosmic-ray ionization rate of ζ ≳ 10−14 s−1 (3–4 orders of magnitude higher than that within clouds in the Galactic spiral arms) to reproduce the observed results. From the difference in formation routes of HOC+, we propose that a low-excitation line of HOC+ traces cosmic-ray dominated regions, while high-excitation lines trace photodissociation regions. Our results suggest that the interstellar medium in the center of NGC 253 is significantly affected by energy input from UV photons and cosmic rays, sources of energy feedback.

Clinical measures of communication limitations in dysarthria assessed through crowdsourcing: specificity, sensitivity, and retest-reliability

ABSTRACT Assessing the impact of dysarthria on a patient’s ability to communicate should be an integral part of patient management. However, due to the high demands on reliable quantification of communication limitations, hardly any formal clinical tests with approved psychometric properties have been developed so far. This study investigates a web-based assessment of communication impairment in dysarthria, named KommPaS. The test comprises measures of intelligibility, naturalness, perceived listener effort and communication efficiency, as well as a total score that integrates these parameters. The approach is characterized by a quasi-random access to a large inventory of test materials and to a large group of naïve listeners, recruited via crowdsourcing. As part of a larger research program to establish the clinical applicability of this new approach, the present paper focuses on two psychometric issues, namely specificity and sensitivity (study 1) and retest-reliability (study 2). Study 1: KommPaS was administered to 54 healthy adults and 100 adult persons with dysarthria (PWD). Non-parametric criterion-based norms (specificity: 0.95) were used to derive a standard metric for each of the four component variables, and corresponding sensitivity values for the presence of dysarthria were identified. Overall classification accuracy of the total score was determined using a ROC analysis. The resulting cutscores showed a high accuracy in the separation of PWD from healthy speakers for the naturalness and the total score. Study 2: A sub-group of 20 PWD enrolled in study 1 were administered a second KommPaS examination. ICC analyses revealed good to excellent retest reliabilities for all parameters.

Interpreting the Old English Nominal Group from a Parsed Corpus

The object of this study is to derive a systemic functional interpretation of the experiential structure of the Old English nominal group from a parsed database. The database is the York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE), a parsed structuralist corpus. The methodology uses the CorpusSearch search engine to isolate nominal group tree-structures in two separate inventories. The first search results in an inventory of the nominal groups having the most extended experiential structures without postmodifiers. A functional interpretation of the experiential structures revealed shows that the categories of experiential elements in the modern English nominal group are enough to explain the Old English structures. The second search results in a much larger inventory of the nominal groups having the next most extended experiential structures without postmodifiers. Structural and functional interpretations of these groups confirm the variety of experiential elements in the first inventory, and show numerous types of variation in element sequence and realizations. They also show the need for an additional element category, Pre-deictic. Variations in sequence include Epithet before Numerative, and Epithet or Numerative before Deictic. Realizational variations include Classifiers realized by embedded nominal groups.

Vowel acoustics in Greek-speaking children with cochlear implants

Children who are deaf have been reported to have atypical, often reduced vowel spaces. Improved auditory input via a cochlear implant [CI] may ameliorate such issues, although wide variation among CI users has been observed. Much of the existing literature on speech development post-cochlear implantation assessed speakers of English, which has a rather large vowel inventory. Smaller vowel inventories may be associated with differences in the overall vowel space area as well as in the size of individual vowel areas. This study investigates children acquiring Greek, which has five phonemic vowels /i ɛ a ɔ u/. Seven children with CIs, ages 4–16 years, were age-matched with typically developing children. Vowels were elicited in disyllabic words matched for preceding (bilabial) and following (alveolar) phonetic context: pita, peta, pata, puse, pote. Ten productions of each word were elicited using a picture-naming paradigm. Analyses will compare overall vowel space sizes and the degree of variability of individual vowel categories between the two groups. Investigations of children learning phonologies with smaller vowel inventories will lead to a fuller picture of how deafness and subsequent cochlear implantation impact speech communication across languages. [Work supported by Fulbright Greece and iCARE.]

A Multispecies Pseudoadiabat for Simulating Condensable-rich Exoplanet Atmospheres

Abstract Central stages in the evolution of rocky, potentially habitable planets may play out under atmospheric conditions with a large inventory of nondilute condensable components. Variations in condensate retention and accompanying changes in local lapse rate may substantially affect planetary climate and surface conditions, but there is currently no general theory to effectively describe such atmospheres. In this article, expanding on the work by Li et al., we generalize the single-component moist pseudoadiabat derivation in Pierrehumbert to allow for multiple condensing components of arbitrary diluteness and retained condensate fraction. The introduction of a freely tunable retained condensate fraction allows for a flexible, self-consistent treatment of atmospheres with nondilute condensable components. To test the pseudoadiabat’s capabilities for simulating a diverse range of climates, we apply the formula to planetary atmospheres with compositions, surface pressures, and temperatures representing important stages with condensable-rich atmospheres in the evolution of terrestrial planets: a magma ocean planet in a runaway greenhouse state; a post-impact, late-veneer-analog planet with a complex atmospheric composition; and an Archean Earth-like planet near the outer edge of the classical circumstellar habitable zone. We find that variations in the retention of multiple nondilute condensable species can significantly affect the lapse rate and in turn outgoing radiation and the spectral signatures of planetary atmospheres. The presented formulation allows for a more comprehensive treatment of the climate evolution of rocky exoplanets and early Earth analogs.

Data–driven classification of landslide types at a national scale by using Artificial Neural Networks

Classification of landslide type is an essential step in risk management, although is often missing in large inventories. Here we propose a novel data-driven method that uses easily accessible morphometric and geospatial input parameters to classify landslides type at a national scale in Italy by means of a shallow Artificial Neural Network. We achieved an overall True Positive Rate of 0.76 for a five-class overall classification of over 275,000 landslides as (1) rockfall/toppling, (2) translational/rotational slide, (3) earth flow, (4) debris flow, and (5) complex landslide. In general, the model performance is very good in the entire national territory, with large areas reaching F-score higher than 0.9. The method can be applied to any polygonal inventory, as those produced by automatic mapping procedures from Earth Observation imagery, in order to automatically identify the types of landslides.

Evaluation of intensity measure performance in regional seismic risk assessment of reinforced concrete bridge inventories

Seismic risk assessment requires fragility functions derived using non-linear time-history analysis with earthquake records compatible with the site hazard and characterised via a suitable intensity measure (IM), which poses a challenge given the wide variety of structural characteristics found in bridge inventories. Given the lack of consensus on a suitable IM, regional studies often use peak ground acceleration (PGA), which despite being recognized as a poor indicator of structural performance, has persisted in earthquake response characterisation. Average spectral acceleration (AvgSa) has recently gained popularity since it describes earthquake intensity over a range of pertinent periods of vibration; however, its suitability as an IM has not been demonstrated on real bridge inventories with a wide variety of structural characteristics. In this study, hazard-consistent records selected for both PGA and AvgSa are used to evaluate the response of a large inventory of existing bridges with varying structural characteristics, from the Italian roadway network. The results indicate that AvgSa is a more accurate IM compared to PGA when applied to the regional assessment of RC bridges.

Designing a fully automated and integrated inventory and replenishment system for hospitals

Surge demand shows that hospital logistics automation is unexploited given that the high pressure in work, the lack of resources, and some inventory and replenishment problems still occur. Hospitals do not fully leverage the implemented automatic systems and integrate them into large-scale logistic processes. Only some tasks/ services are automated keeping nurses in charge of the logistic labour, a large stock space, and inventory and replenishment problems. Hospitals do not adopt strategies that carefully aim for a high level of logistics automation and integration, using adequate management systems/ policies to make it cost-effective. This paper presents the design of a fully automated and integrated system for asset inventory and replenishment management in hospitals. The strategy adopted in this work starts with choosing a proactive replenishment policy operating in real-time and a suitable inventory and replenishment system (RFID-enabled two-bin system). This system is then automated using adequate and recent technologies (e.g. ‘Box-Picker’, smart gravity-flow racks, and conveyors). The proposed system is modelled and simulated in a virtual hospital under a mass casualty incident. Results show a considerable reduction of assets unavailability (99.98% in the central pharmacy and 90.47% in wards) and stock space. Besides, nurses are released from logistic tasks.

Using Machine Learning to Parse Chemical Mixture Descriptions

Chemical mixtures have recently come to the attention of open standards and data structures for capturing machine-readable descriptions for informatics uses. At the present time, essentially all transmission of information about mixtures is done using short text descriptions that are readable only by trained scientists, and there are no accessible repositories of marked-up mixture data. We have designed a machine learning tool that can interpret mixture descriptions and upgrade them to the high-level Mixfile format, which can in turn be used to generate Mixtures InChI notation. The interpretation achieves a high success rate and can be used at scale to markup large catalogs and inventories, with some expert checking to catch edge cases. The training data that was accumulated during the project is made openly available, along with previously released mixture editing tools and utilities.

Improving broad-coverage medical entity linking with semantic type prediction and large-scale datasets

ObjectivesBiomedical natural language processing tools are increasingly being applied for broad-coverage information extraction—extracting medical information of all types in a scientific document or a clinical note. In such broad-coverage settings, linking mentions of medical concepts to standardized vocabularies requires choosing the best candidate concepts from large inventories covering dozens of types. This study presents a novel semantic type prediction module for biomedical NLP pipelines and two automatically-constructed, large-scale datasets with broad coverage of semantic types. MethodsWe experiment with five off-the-shelf biomedical NLP toolkits on four benchmark datasets for medical information extraction from scientific literature and clinical notes. All toolkits adopt a staged approach of mention detection followed by two stages of medical entity linking: (1) generating a list of candidate concepts, and (2) picking the best concept among them. We introduce a semantic type prediction module to alleviate the problem of overgeneration of candidate concepts by filtering out irrelevant candidate concepts based on the predicted semantic type of a mention. We present MedType, a fully modular semantic type prediction model which we integrate into the existing NLP toolkits. To address the dearth of broad-coverage training data for medical information extraction, we further present WikiMed and PubMedDS, two large-scale datasets for medical entity linking. ResultsSemantic type filtering improves medical entity linking performance across all toolkits and datasets, often by several percentage points of F-1. Further, pretraining MedType on our novel datasets achieves state-of-the-art performance for semantic type prediction in biomedical text. ConclusionsSemantic type prediction is a key part of building accurate NLP pipelines for broad-coverage information extraction from biomedical text. We make our source code and novel datasets publicly available to foster reproducible research.

Low-Cost Asset Tracking and Inventory Management System (ATIM)

In any supply chain, the warehouse is the main component in linking the chain partners and nowadays it acts as a competitive factor. Current Inventory management Systems are slow, inefficient concerning managing and tracking the huge demand for large inventory traffic and hence make a poor customer experience while managing large, urgent and important inventories. Internet of Things (IoT) is a promising technology that can be used in the context of Industry 4.0. This paper aims to improve inventory management efficiency multi-folds and enhance stock SCM (supply chain management) analysis for market planning. This solution strengthens store operations & inventory management smartly, safely and efficiently. Also, the blocks and layers of IoT are illustrated. Finally, the benefits and challenges of implementing IoT in the SCM and the warehouse are discussed.

Fuzzy‐based method for efficient seismic performance evaluation of structures with uncertainty

AbstractSeismic performance evaluation of a large inventory of old structures with no design document is a time‐consuming and costly task. In this research, a fuzzy‐based method and its toolbox are developed to efficiently approximate the seismic performance of structures with lack of structural information. The uncertain parameters are approximated as fuzzy numbers based on statistical studies on existing buildings, empirical formulas, alternative analysis methods, visual assessment, and design codes. The statistical database provided in this research can be valuable to make approximation for seismic performance of structures. The accuracy and applicability of the developed method are demonstrated using two real structures. Furthermore, the applicability of this study is demonstrated using the sensitivity and life‐cycle cost analysis of the case study structures. It is observed that the proposed method can approximate the seismic performance of a structure in a considerably short time with an acceptable trade‐off compared with more costly and time‐consuming site evaluations and detailed computations.

Topography and geology effects on travel distances of natural terrain landslides: Evidence from a large multi-temporal landslide inventory in Hong Kong

Landslides pose threats to the safety and property of people living in mountainous areas like Hong Kong. Dependence of landslide travel distance on initial conditions and travel routes has broad implications for assessing the landslide risks. In this study, the controlling topographic and geologic factors for travel distances of open hillslope landslides and channelized debris flows are examined based on the Enhanced Natural Terrain Landslide Inventory (ENTLI) in Hong Kong with 11,622 records during the period of 1984–2013. The mean values of the horizontal travel distances of open hillslope landslides and channelized debris flows are 24.1 and 86.3 m, respectively. Correlations combining horizontal travel distances, volumes and fall heights of open hillslope landslides have been derived. The horizontal travel distances (L) and fall heights (H) are closely related for both types of landslides. The use of the L/H ratio alone as a representation of natural terrain landslide mobility is not suggested. Because of the entrainment of loose materials, the travel distances of channelized debris flows are enlarged significantly on colluvium lands (with mean value of 142.6 m), compared with those on weathered mantles (with mean value of 74.2 m).

Deep Neural Networks with Transfer Learning for Forest Variable Estimation Using Sentinel-2 Imagery in Boreal Forest

Estimation of forest structural variables is essential to provide relevant insights for public and private stakeholders in forestry and environmental sectors. Airborne light detection and ranging (LiDAR) enables accurate forest inventory, but it is expensive for large area analyses. Continuously increasing volume of open Earth Observation (EO) imagery from high-resolution (&lt;30 m) satellites together with modern machine learning algorithms provide new prospects for spaceborne large area forest inventory. In this study, we investigated the capability of Sentinel-2 (S2) image and metadata, topography data, and canopy height model (CHM), as well as their combinations, to predict growing stock volume with deep neural networks (DNN) in four forestry districts in Central Finland. We focused on investigating the relevance of different input features, the effect of DNN depth, the amount of training data, and the size of image data sampling window to model prediction performance. We also studied model transfer between different silvicultural districts in Finland, with the objective to minimize the amount of new field data needed. We used forest inventory data provided by the Finnish Forest Centre for model training and performance evaluation. Leaving out CHM features, the model using RGB and NIR bands, the imaging and sun angles, and topography features as additional predictive variables obtained the best plot level accuracy (RMSE% = 42.6%, |BIAS%| = 0.8%). We found 3×3 pixels to be the optimal size for the sampling window, and two to three hidden layer DNNs to produce the best results with relatively small improvement to single hidden layer networks. Including CHM features with S2 data and additional features led to reduced relative RMSE (RMSE% = 28.6–30.7%) but increased the absolute value of relative bias (|BIAS%| = 0.9–4.0%). Transfer learning was found to be beneficial mainly with training data sets containing less than 250 field plots. The performance differences of DNN and random forest models were marginal. Our results contribute to improved structural variable estimation performance in boreal forests with the proposed image sampling and input feature concept.