Hellenic Petroleum (Greece)

Charts are very popular for analyzing data.When exploring charts, people often ask a variety of complex reasoning questions that involve several logical and arithmetic operations.They also commonly refer to visual features of a chart in their questions.However, most existing datasets do not focus on such complex reasoning questions as their questions are template-based and answers come from a fixedvocabulary.In this work, we present a largescale benchmark covering 9.6K human-written questions as well as 23.1K questions generated from human-written chart summaries.To address the unique challenges in our benchmark involving visual and logical reasoning over charts, we present two transformer-based models that combine visual features and the data table of the chart in a unified way to answer questions.While our models achieve the state-of-the-art results on the previous datasets as well as on our benchmark, the evaluation also reveals several challenges in answering complex reasoning questions.

Maarten Sap, Swabha Swayamdipta, Laura Vianna, Xuhui Zhou, Yejin Choi, Noah Smith. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Aspect term extraction (ATE) aims at identifying all aspect terms in a sentence and is usually modeled as a sequence labeling problem.However, sequence labeling based methods cannot make full use of the overall meaning of the whole sentence and have the limitation in processing dependencies between labels.To tackle these problems, we first explore to formalize ATE as a sequence-tosequence (Seq2Seq) learning task where the source sequence and target sequence are composed of words and labels respectively.At the same time, to make Seq2Seq learning suit to ATE where labels correspond to words one by one, we design the gated unit networks to incorporate corresponding word representation into the decoder, and position-aware attention to pay more attention to the adjacent words of a target word.The experimental results on two datasets show that Seq2Seq learning is effective in ATE accompanied with our proposed gated unit networks and position-aware attention mechanism.

Integration of a process-based crop simulation model with user-friendly expert systems has aided farm managers by facilitating the selection of optimal solutions to widely varying problems. As such systems are enhanced to further understand plant responses to environment, there is increased need for diagnostics and management-decision aids either in support of optimizing resources for efficient farm management in precision agriculture technologies, or global climate change research, or the use of plants for remediation of extreme environmental conditions. In regards to precision agriculture, most engineering and computing technologies are presently in reality or commercially available for variable-rate/site-specific management; whereas, the application of crop simulation models has been hampered by a lack of understanding of responses of several key physiological and developmental processes to environmental variation and the failure of many to conceptualize the opportunities to apply such technology to real-world agricultural and environmental problems. There are certainly a variety of approaches and facilities for investigating plant response to the environment. We have demonstrated the utility and value of a Soil-Plant-Atmosphere-Research (SPAR) facility, which comprises ten outdoor, naturally-lit chambers, in generating data useful for increased understanding of cotton growth and physiological responses to environment and for developing process-level physiological models. Operating a SPAR facility to acquire model data is often being more expedient and economical than field-plot experiments, because SPAR allows the scientist to minimize many of the covarying and confounding factors that occur in field experiments. As a result, basic plant processes investigated can be more directly related to the environmental variable(s) being studied. Also, the SPAR facilities are optimized for the measurement of plant and canopy-level physiological, growth and developmental processes under precisely controlled, but naturally lit, environmental conditions. This paper presents operational data and research results from a SPAR facility at Mississippi State University, Mississippi State, constructed in 1977 and still in use today. We describe herein how data obtained in the past and as well as data in future studies have features that are unique and instructive for both basic and applied plant biologists.

We use machine learning techniques to find the best combination of local focus and lexical distance features for identifying the anchor of mereological bridging references.We find that using first mention, utterance distance, and lexical distance computed using either Google or WordNet results in an accuracy significantly higher than obtained in previous experiments.

Yujia Qin, Yankai Lin, Ryuichi Takanobu, Zhiyuan Liu, Peng Li, Heng Ji, Minlie Huang, Maosong Sun, Jie Zhou. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Abstract Geophysical and geological data from the eastern sector of the Central European Variscan belt are presented and reviewed in the regional tectonic context. Matched filtering of isostatic gravity, guided by results of spectral analysis, along with other derivatives of gravity and magnetic fields reveal a dominant WNW–ESE-trending pre-Permian structural grain in the external zones of the Variscan belt in Poland. This trend is confirmed by regional distribution of dips in Carboniferous and Devonian strata that were penetrated by boreholes beneath Permian-Mesozoic sediments. Based on these data, two alternative concepts explaining the connection of the Variscan belt and its NE foreland, those of strike-slip tectonics versus oroclinal bending, are discussed. The WNW–ESE structural trend in the Variscan foreland is parallel to a set of major strike-slip fault zones in the area, including those of Upper Elbe, Intra-Sudetic, Odra, Dolsk and Kraków-Lubliniec. These faults are considered to convey a significant dextral displacement between Laurussia and Gondwana. The revised position of the Variscan deformation front shows a similar, uninterrupted, generally WNW–ESE trend, up to the SE border of Poland, which indicates an initial continuation of the Variscan belt into the area of the present-day Western Carpathians. The geometry of the Variscan deformation front along with the pattern of the Variscan structural grain are inconsistent with the idea of an oroclinal loop affecting the external, non-metamorphic Variscan belt. However, the data presented do not entirely rule out an oroclinal loop within the Variscan internides. The still possible options are (1) a semi-oroclinal model postulating ~ 90° bending of the Variscan tectonostratigraphic zones into parallelism with the WNW–ESE strike-slip faults or (2) an orocline limited only to the belt linking the Wolsztyn High and Moravo-Silesian non- to weakly-metamorphic fold-and-thrust belt. Regardless of the kinematic model preferred, our data indicate that structural evolution of the Polish Variscides was concluded with the end-Carboniferous NNE–SSW shortening that resulted in the present-day extent of the Variscan deformation front.

This paper reviews previous data on the geological evolution of Western Greece, with special emphasis on the petroleum potential of the Pre‐Apulian zone (including new data) and the Ionian zone, the two most external portions of the Hellenide fold‐and‐thrust belt. From the Triassic to the Late Cretaceous, Western Greece constituted part of the southern passive margin of Tethys, and siliceous facies are widely associated with organic‐carbon rich deposits. Pelagic Late Jurassic units rich in marine organic matter constitute important hydrocarbon source rocks in the pelagic‐neritic Pre‐Apulian zone succession. Oil‐oil correlation with an Apulian zone oil sample (from Aquila, Italy) indicates similar geochemical characteristics. Thus, the significant volumes of oil generated by the rich and mature source rock intervals identified in the Pre‐Apulian zone are likewise expected to be of good quality. In the Ionian zone, four organic‐carbon rich intervals with hydrocarbon potential have been recorded. The tectonic history of the Pre‐Apulian zone, which is characterised by the presence of large anticlines, is favourable for the formation of structural traps. By contrast, locations suitable for the entrapment of hydrocarbons in the Ionian zone are restricted to small anticlines within largerscale synclinal structures. Hydrocarbon traps may potentially be present at the tectonic contacts between the Ionian zone and both the Pre‐Apulian and Gavrovo zones. Major traps may also have been formed between the pre‐evaporitic basement and the evaporite‐dominated units at the base of both the Pre‐Apulian and the Ionian zone successions. The degree of participation of the subevaporitic basement in the deformation of the Pre‐Apulian and Ionian sedimentary cover will determine the location and size of these traps. Various scenarios regarding the deformation of the sub‐evaporitic succession are examined in order to determine the hydrocarbon trapping possibilities of each model. The hypothesis of continental subduction (Early to Late Miocene) of the shared pre‐evaporitic basement of the Pre‐Apulian and Ionian zone eastwards of the Ionian zone is regarded favourably, as it appears to be compatible with the presence of a Phyllite ‐ Quartzite ‐ dominated (HP/LT) metamorphic unit beneath the Gavrovo‐Tripolis zone carbonates in Peloponnesus and Crete.

Contractional structures recognised in a recent SW‐NE oriented seismic profile offshore western Greece, between the islands of Zakynthos and Kefallinia (Cephalonia), indicate that this part of the Pre‐Apulian geotectonic zone was involved in Quaternary shortening related to the westward propagation of the Hellenic fold‐and‐thrust system. Deep reflector horizons including the Moho and the top of the crystalline basement were identified on the profile. Shallower reflectors include those corresponding to the contacts between the Mesozoic/Miocene, Upper Miocene/Lower Pliocene, and Pliocene/Pleistocene sedimentary sequences. The Upper Cenozoic to Quaternary sequence rests unconformably upon Mesozoic carbonates. Triassic evaporites wedge‐out in the Paxos geotectonic zone, where the Palaeozoic passes up into Mesozoic deposits. We have identified contractional structures which were reactivated during the Plio‐Quaternary on pre‐existing high‐angle normal faults, and which gave rise to significant topographic anomalies. West‐dipping normal faults were also recognised both within the Palaeozoic and Cenozoic successions, and are related to regional extension during sedimentation. East‐dipping thrust faults which root in the evaporites were also identified on the seismic profile. Due to right‐lateral strike‐slip activity on the Kefallinia Transform Fault, east‐dipping normal faults were formed within the lonion abyssal plain. This abyssal plain together with the Hellenic Trench, an accretionary prism, and a forearc basin can be recognised on the seismic profile. A “triple junction” between the Apulian (African) Platform, the oceanic crust of the Ionian Abyssal Plain and the Eurasian Plate in the west of the line is related to the Kefallinia Transform Fault. Neotectonic structural deformation (i.e. Quaternary‐Holocene) is superimposed on the above‐mentioned structures. Finally, diapiric movement of Triassic evaporites has affected both the Alpine and the late Cenozoic to Holocene sedimentary sequences. Diapiric activity continues at the present day in the eastern part of the profile, in the lonian geotectonic zone. The forearc basin may be prospective for hydrocarbons. Target areas include the lonian channel where a play has already been located, and its extension to the south (the Kyparisiakos Gulf area). Here, thick late Cenozoic to Quaternary deposits may act as a top‐seal above a reservoir consisting of eroded Mesozoic to Eocene carbonates, as at the recent Katakolon discovery.

The model of a foreland propagating sequence already presented for the External Hellenides is significantly modified in this paper. New data are used, including structural maps, cross-sections, stratigraphic determinations and seismic profiles. In general, thrusts formed a foreland propagating sequence but they acted simultaneously for a long period of time. Thus, during the Middle Eocene the Pindos thrust resulted in the formation of the Ionian–Gavrovo foreland and acted in tandem with the newly formed Gavrovo thrust within the basin until the Late Oligocene. The Gavrovo thrust consists of segments, showing that out-of-sequence thrusting was important. Thrust nucleation and propagation history is strongly influenced by normal faults formed in the forebulge region of the Ionian–Gavrovo foreland basin. Shortening rates within the Gavrovo–Ionian foreland are low, about 1 mm/year. Although thrust load played an important role in the formation of this basin, the additional load of 3500 m thick clastics in the basin enhanced subsidence and underthrusting.

Ankur Parikh, Xuezhi Wang, Sebastian Gehrmann, Manaal Faruqui, Bhuwan Dhingra, Diyi Yang, Dipanjan Das. Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP). 2020.

QA TempEval shifts the goal of previous TempEvals away from an intrinsic evaluation methodology toward a more extrinsic goal of question answering.This evaluation requires systems to capture temporal information relevant to perform an end-user task, as opposed to corpus-based evaluation where all temporal information is equally important.Evaluation results show that the best automated TimeML annotations reach over 30% recall on questions with 'yes' answer and about 50% on easier questions with 'no' answers.Features that helped achieve better results are event coreference and a time expression reasoner.

Charts are widely used for data analysis, providing visual representations and insights into complex data.To facilitate chart-based data analysis using natural language, several downstream tasks have been introduced recently such as chart question answering and chart summarization.However, existing methods for these tasks often rely on pretraining on language or vision-language tasks, neglecting the explicit modeling of chart structures (e.g., how chart elements are related to each other).To address this, we first build a large corpus of charts covering diverse topics and visual styles.We then present UniChart, a pretrained model for chart comprehension and reasoning.UniChart encodes the relevant text, data, and visual elements of charts and then uses a chart-grounded text decoder for text generation.We propose several chart-specific pretraining tasks that include: (i) low-level tasks to extract the visual elements (e.g., bars, lines) and data from charts, and (ii) high-level tasks to acquire chart understanding and reasoning skills.Our experiments demonstrate that pretraining UniChart on a large corpus with chart-specific objectives, followed by fine-tuning, yields state-of-the-art performance on four downstream tasks.Moreover, our model exhibits superior generalizability to unseen chart corpus, surpassing previous approaches that lack chart-specific objectives and utilize limited chart resources.

Dialogue models trained on human conversations inadvertently learn to generate toxic responses.In addition to producing explicitly offensive utterances, these models can also implicitly insult a group or individual by aligning themselves with an offensive statement.To better understand the dynamics of contextually offensive language, we investigate the stance of dialogue model responses in offensive Reddit conversations.Specifically, we create TOXICHAT, a crowd-annotated dataset of 2,000 Reddit threads and model responses labeled with offensive language and stance.Our analysis reveals that 42% of human responses agree with toxic comments, whereas only 13% agree with safe comments.This undesirable behavior is learned by neural dialogue models, such as DialoGPT, which we show are two times more likely to agree with offensive comments.To enable automatic detection of offensive language, we fine-tuned transformerbased classifiers on TOXICHAT that achieve 0.71 F 1 for offensive labels and 0.53 Macro-F 1 for stance labels.Finally, we quantify the effectiveness of controllable text generation (CTG) methods to mitigate the tendency of neural dialogue models to agree with offensive comments.Compared to the baseline, our best CTG model achieves a 19% reduction in agreement with offensive comments and produces 29% fewer offensive replies.Our work highlights the need for further efforts to characterize and analyze inappropriate behavior in dialogue models, in order to help make them safer. 1

Evaluating the quality of generated text is a challenging task in NLP, due to the inherent complexity and diversity of text.Recently, large language models (LLMs) have garnered significant attention due to their impressive performance in various tasks.Therefore, we present this paper to investigate the effectiveness of LLMs, especially ChatGPT, and explore ways to optimize their use in assessing text quality.We compared three kinds of referencefree evaluation methods.The experimental results prove that ChatGPT is capable of evaluating text quality effectively from various perspectives without reference and demonstrates superior performance than most existing automatic metrics.In particular, the Explicit Score, which utilizes ChatGPT to generate a numeric score measuring text quality, is the most effective and reliable method among the three exploited approaches.However, directly comparing the quality of two texts may lead to suboptimal results.We believe this paper will provide valuable insights for evaluating text quality with LLMs and have released the used data 1 .

Cosmological and astrophysical observations suggest that 85% of the total matter of the Universe is made of Dark Matter (DM). However, its nature remains one of the most challenging and fundamental open questions of particle physics. Assuming particle DM, this exotic form of matter cannot consist of Standard Model (SM) particles. Many models have been developed to attempt unraveling the nature of DM such as Weakly Interacting Massive Particles (WIMPs), the most favored particle candidates. WIMP annihilations and decay could produce SM particles which in turn hadronize and decay to give SM secondaries such as high energy 𝛾 rays. In the framework of indirect DM search, observations of promising targets are used to search for signatures of DM annihilation. Among these, the dwarf spheroidal galaxies (dSphs) are commonly favored owing to their expected high DM content and negligible astrophysical background. In this work, we present the very first combination of 20 dSph observations, performed by the Fermi-LAT, HAWC, H.E.S.S., MAGIC, and VERITAS collaborations in order to maximize the sensitivity of DM searches and improve the current results. We use a joint maximum likelihood approach combining each experiment’s individual analysis to derive more constraining upper limits on the WIMP DM self-annihilation cross-section as a function of DM particle mass. We present new DM constraints over the widest mass range ever reported, extending from 5 GeV to 100 TeV thanks to the combination of these five different 𝛾-ray instruments.

Results of a depth-to-basement study are presented for the East European Craton and the Teisseyre-Tornquist Zone (TTZ) in Poland. The terrestrial gravity data are inverted for the top of the Paleoproterozoic basement and, independently, for the top of the Ediacaran using seismic horizons from the PolandSPAN™ seismic survey and well tops as input depth measurements. The depth to the Ediacaran modelling was additionally extended to cover the Łysogóry Block and northern Małopolska Block. The results are visualised as isobath maps for the top of the Paleoproterozoic basement and top of the Ediacaran and an isopach map for the Ediacaran, supplemented with qualitative structural interpretation based on gravity and magnetic data. The results of modelling show a smooth crystalline basement slope within the TTZ with the top of the Paleoproterozoic basement uniformly descending south-westwards by 10–14 km. The thickness of the Ediacaran in SE Poland increases in the same direction to more than 10 km within the TTZ. Such a crustal architecture, in combination with the earlier documented Moho elevation of 4–6 km, reveals significant thinning of the Paleoproterozoic crust within the TTZ to form a crustal necking zone due to the Ediacaran rifting. A smooth geometry of the top of basement along with the lack of basement-rooted faults suggests a ductile mode of crustal thinning during rifting of Rodinia. Moreover, the development of the NW–SE-oriented rift, a precursor of the Tornquist Ocean, was associated with rifting in a NE–SW direction parallel to the Orsha-Volyn Rift.

There have been many attempts at automatically recognising author personality traits from text, typically incorporating linguistic features with conventional machine learning models, e.g.linear regression or Support Vector Machines.In this work, we propose to use deep-learningbased models with atomic features of text -the characters -to build hierarchical, vectorial word and sentence representations for the task of trait inference.On a corpus of tweets, this method shows stateof-the-art performance across five traits and three languages (English, Spanish and Italian) compared with prior work in author profiling.The results, supported by preliminary visualisation work, are encouraging for the ability to detect complex human traits.

Abstract The Tajik basin and southwestern Tian Shan constitute the northwestern tip of the India‐Asia collision zone. Basin inversion formed the thin‐skinned Tajik fold‐thrust belt, outlined by westward convex fold trains, underlain by a décollement in Jurassic evaporites. The belt's leading edge—the Uzbek Gissar—and its transpressional northern lateral margin—the Tajik Gissar—constitute the thick‐skinned foreland buttresses. Apatite fission‐track data indicate ~40‐ to 15‐Ma reheating by sediment burial in the Tian Shan. In the Gissar and the Tajik fold‐thrust belt, apatite fission‐track and (U,Th)/He ages date the major phase of shortening/erosion between ~12 and 1 Ma, with exhumation to 2‐ to 3‐km crustal depths within a few Myr after onset of shortening. Shortening spread immediately across the fold‐thrust belt, typical for belts floored by a detachment in ductile rocks, and into the foreland buttresses. Reactivation concentrated in the internal (eastern) fold‐thrust belt with the thickest evaporates. The youngest ages (~6.6–1.6 Ma) occur along the Vakhsh thrust, the active erosional front of the fold‐thrust belt in the northeastern Tajik basin, where it narrows between the converging Tian Shan and Pamir. Our study links major events in the Pamir hinterland with the Tajik basin and Tian Shan foreland. In the late Eocene–early Miocene, the advancing Pamir‐plateau crust loaded the foreland, inducing subsidence, reheating, and early shortening. Basin inversion and major shortening/transpression in the foreland buttresses from ~12 Ma onward were synchronous with the subcrustal indentation of Indian lithosphere into the Tajik‐Tarim basin lithosphere and the onset of its rollback beneath the Pamir.

Research Article| July 01, 2012 Structural decoupling in a convergent forearc setting (southern Crete, Eastern Mediterranean) Eleni Kokinou; Eleni Kokinou † 1Technological Educational Institute Crete, 3 Romanou Street Chalepa, Chania, Crete, GR 73133 Greece †E-mail: ekokinou@chania.teicrete.gr Search for other works by this author on: GSW Google Scholar Alves Tiago; Alves Tiago 23D Seismic Laboratory, School of Earth and Ocean Sciences, Cardiff University, Main Building-Park Place, CF10 3AT Cardiff, UK Search for other works by this author on: GSW Google Scholar Kamberis Evangelos Kamberis Evangelos 3Hellenic Petroleum (Exploration and Production of Hydrocarbons Division), 199 Kifissias Avenue, 15124 Maroussi, Athens, Greece Search for other works by this author on: GSW Google Scholar Author and Article Information Eleni Kokinou † 1Technological Educational Institute Crete, 3 Romanou Street Chalepa, Chania, Crete, GR 73133 Greece Alves Tiago 23D Seismic Laboratory, School of Earth and Ocean Sciences, Cardiff University, Main Building-Park Place, CF10 3AT Cardiff, UK Kamberis Evangelos 3Hellenic Petroleum (Exploration and Production of Hydrocarbons Division), 199 Kifissias Avenue, 15124 Maroussi, Athens, Greece †E-mail: ekokinou@chania.teicrete.gr Publisher: Geological Society of America Received: 13 Feb 2011 Revision Received: 07 Dec 2011 Accepted: 08 Dec 2011 First Online: 08 Mar 2017 Online ISSN: 1943-2674 Print ISSN: 0016-7606 © 2012 Geological Society of America GSA Bulletin (2012) 124 (7-8): 1352–1364. https://doi.org/10.1130/B30492.1 Article history Received: 13 Feb 2011 Revision Received: 07 Dec 2011 Accepted: 08 Dec 2011 First Online: 08 Mar 2017 Cite View This Citation Add to Citation Manager Share Icon Share Facebook Twitter LinkedIn Email Permissions Search Site Citation Eleni Kokinou, Alves Tiago, Kamberis Evangelos; Structural decoupling in a convergent forearc setting (southern Crete, Eastern Mediterranean). GSA Bulletin 2012;; 124 (7-8): 1352–1364. doi: https://doi.org/10.1130/B30492.1 Download citation file: Ris (Zotero) Refmanager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex toolbar search Search Dropdown Menu toolbar search search input Search input auto suggest filter your search All ContentBy SocietyGSA Bulletin Search Advanced Search Abstract A multidisciplinary approach is used to investigate the structure of the southern Cretan margin, which is located in one of the most seismically active forearc regions in Europe. Bathymetric, seismic-reflection, and fault plane solution data were used to identify the main tectonic features on the margin, correlating their evolution with the main sedimentary sequences recognized on Crete. In contrast to the majority of forearc settings in the Pacific and Indian Oceans, southern Crete is a region of predominantly oblique movement above well-defined detachment zones. North-dipping thrust faults identified on seismic-reflection profiles reveal significant crustal shortening during the Miocene due to the westward propagation of the Hellenic fold-and-thrust system. In addition, east-dipping thrust faults rooted on top of pre-Neogene strata were also identified, but only a few of these thrusts affect Neogene to Holocene strata. Small-scale domes derived from evaporitic (Messinian) intrusions deform Pliocene–Quaternary strata. West- and east-dipping normal faults were also recognized within the Mesozoic and Cenozoic successions, and these are related to regional extension during forearc convergence. In such a setting, the fault-bounded continental slope of Crete effectively separates a region of uplift (Crete) from subsiding troughs to the south. Our work shows that structural segmentation at depth is complex, with multiple crustal levels recording contrasting styles of deformation and distinct moment-tensor solutions. This complexity derives from the oblique style of convergence recorded south of Crete, which reactivates distinct crustal levels depending on their rheology and relative degree of metamorphism inherited during Alpine compression. As a result, a strong correlation between seafloor morphology and transtensional movements is recorded in the upper 10–15 km of the crust, while transpression prevailed after the Serravallian below these depths. You do not have access to this content, please speak to your institutional administrator if you feel you should have access.