LkeRec: Toward Lightweight End-to-End Joint Representation Learning for Building Accurate and Effective Recommendation

Explicit and implicit knowledge about users and items have been used to describe complex and heterogeneous side information for recommender systems (RSs). Many existing methods use knowledge graph embedding (KGE) to learn the representation of a user-item knowledge graph (KG) in low-dimensional space. In this article, we propose a lightweight end-to-end joint learning framework for fusing the tasks of KGE and RSs at the model level. Our method proposes a lightweight KG embedding method by using bidirectional bijection relation-type modeling to enable scalability for large graphs while using self-adaptive negative sampling to optimize negative sample generating. Our method further generates the integrated views for users and items based on relation-types to explicitly model users’ preferences and items’ features, respectively. Finally, we add virtual “recommendation” relations between the integrated views of users and items to model the preferences of users on items, seamlessly integrating RS with user-item KG over a unified graph. Experimental results on multiple datasets and benchmarks show that our method can achieve a better accuracy of recommendation compared with existing state-of-the-art methods. Complexity and runtime analysis suggests that our method can gain a lower time and space complexity than most of existing methods and improve scalability.

Inter-ecosystem relation (an attempt of classification)

The importance of inter-ecosystem relations is a priori clear to most of ecologists. Inter-ecosystem relations provide for the integrity of ecological systems of all levels (from consortion to biosphere). The relations between consortions and other ecosystems have not been studied in detail so far. Hence, there is a lack of information about classification of inter-ecosystem relations. Among the most well-studied are anthropogenic relations between natural ecosystems and urboecosystems (Holubets, 2000). We propose a classification of the inter-ecosystem relations based on the following criteria: by origin (natural, anthropogenic, natural-anthropogenic), by frequency (dispo­sable, cyclic, acyclic, permanent), by consequences (normal, catastrophic, evolutio­nary), by relation type (trophic, topic, fabric, foric, mediopatic, behavioral). As the proposed classification is a pioneer one, we believe it will be developed and transformed during the field research of the inter-ecosystem relations. A detailed program of research should be elaborated.

Divisors of expected Jacobian type

Divisors whose Jacobian ideal is of linear type have received a lot of attention recently because of its connections with the theory of $D$-modules. In this work we are interested on divisors of expected Jacobian type, that is, divisors whose gradient ideal is of linear type and the relation type of its Jacobian ideal coincides with the reduction number with respect to the gradient ideal plus one. We provide conditions in order to be able to describe precisely the equations of the Rees algebra of the Jacobian ideal. We also relate the relation type of the Jacobian ideal to some $D$-module theoretic invariant given by the degree of the Kashiwara operator.

Task-Driven Knowledge Graph Filtering Improves Prioritizing Drugs for Repurposing

Background Drug repurposing aims at finding new targets for already developed drugs. It becomes more relevant as the cost of discovering new drugs steadily increases. To find new potential targets for a drug, an abundance of methods and existing biomedical knowledge from different domains can be leveraged. Recently, knowledge graphs have emerged in the biomedical domain that integrate information about genes, drugs, diseases and other biological domains. Knowledge graphs can be used to predict new connections between compounds and diseases, leveraging the interconnected biomedical data around them. While real world use cases such as drug repurposing are only interested in one specific relation type, widely used knowledge graph embedding models simultaneously optimize over all relation types in the graph. This can lead the models to underfit the data that is most relevant for the desired relation type. We propose a method that leverages domain knowledge in the form of metapaths and use them to filter two biomedical knowledge graphs (Hetionet and DRKG) for the purpose of improving performance on the prediction task of drug repurposing while simultaneously increasing computational efficiency. Results We find that our method reduces the number of entities by 60% on Hetionet and 26% on DRKG, while leading to an improvement in prediction performance of up to 40.8% on Hetionet and 12.4% on DRKG, with an average improvement of 17.5% on Hetionet and 5.1% on DRKG. Additionally, prioritization of antiviral compounds for SARS CoV-2 improves after task-driven filtering is applied. Conclusion Knowledge graphs contain facts that are counter productive for specific tasks, in our case drug repurposing. We also demonstrate that these facts can be removed, resulting in an improved performance in that task and a more efficient learning process.

LAPREL: A Label-Aware Parallel Network for Relation Extraction

Relation extraction is a crucial task in natural language processing (NLP) that aims to extract all relational triples from a given sentence. Extracting overlapping relational triples from complex texts is challenging and has received extensive research attention. Most existing methods are based on cascade models and employ language models to transform the given sentence into vectorized representations. The cascaded structure can cause exposure bias issue; however, the vectorized representation of each sentence needs to be closely related to the relation extraction with pre-defined relation types. In this paper, we propose a label-aware parallel network (LAPREL) for relation extraction. To solve the exposure bias issue, we apply a parallel network, instead of the cascade framework, based on the table-filling method with a symmetric relation pair tagger. To obtain task-related sentence embedding, we embed the prior label information into the token embedding and adjust the sentence embedding for each relation type. The proposed method can also effectively deal with overlapping relational triples. Compared with 10 baselines, extensive experiments are conducted on two public datasets to verify the performance of our proposed network. The experimental results show that LAPREL outperforms the 10 baselines in extracting relational triples from complex text.

Learning Pattern Relation-Based Hyperbolic Embedding for Adverse Drug Reaction Extraction

Hyperbolic embedding has been recently developed to allow us to embed words in a Cartesian product of hyperbolic spaces, and its efficiency has been proved in several works of literature since the hierarchical structure is the natural form of texts. Such a hierarchical structure exhibits not only the syntactic structure but also semantic representation. This paper presents an approach to learn meaningful patterns by hyperbolic embedding and then extract adverse drug reactions from electronic medical records. In the experiments, the public source of data from MIMIC-III (Medical Information Mart for Intensive Care III) with over 58,000 observed hospital admissions of the brief hospital course section is used, and the result shows that the approach can construct a set of efficient word embeddings and also retrieve texts of the same relation type with the input. With the Poincaré embeddings model and its vector sum (PC-S), the authors obtain up to 82.3% in the precision at ten, 85.7% in the mean average precision, and 93.6% in the normalized discounted cumulative gain.

Information Extraction Tasks based on BERT and SpaCy on Tourism Domain

In this paper, we present two methodologies to extract particular information based on the full text returned from the search engine to facilitate the users. The approaches are based three tasks: name entity recognition (NER), text classification and text summarization. The first step is the building training data and data cleansing. We consider tourism domain such as restaurant, hotels, shopping and tourism data set crawling from the websites. First, the tourism data are gathered and the vocabularies are built. Several minor steps include sentence extraction, relation and name entity extraction for tagging purpose. These steps are needed for creating proper training data. Then, the recognition model of a given entity type can be built. From the experiments, given review texts, we demonstrate to build the model to extract the desired entity,i.e, name, location, facility as well as relation type, classify the reviews or summarize the reviews. Two tools, SpaCy and BERT, are used to compare the performance of these tasks.

Star-triangle type relations from 2d $$ \mathcal{N} $$ = (0, 2) USp(2N) dualities

Inspired by the gauge/YBE correspondence this paper derives some star-triangle type relations from dualities in 2d$$ \mathcal{N} $$ N = (0, 2) USp(2N) supersymmetric quiver gauge theories. To be precise, we study two cases. The first case is the Intriligator-Pouliot duality in 2d$$ \mathcal{N} $$ N = (0, 2) USp(2N) theories. The description is performed explicitly for N = 1, 2, 3, 4, 5 and also for N = 3k + 2, which generalizes the situation in N = 2, 5. For N = 1 a triangle identity is obtained. For N = 2, 5 it is found that the realization of duality implies slight variations of a star-triangle relation type (STR type). The values N = 3, 4 are associated to a similar version of the asymmetric STR. The second case is a new duality for 2d$$ \mathcal{N} $$ N = (0, 2) USp(2N) theories with matter in the antisymmetric tensor representation that arises from dimensional reduction of 4d$$ \mathcal{N} $$ N = 1 USp(2N) Csáki-Skiba-Schmaltz duality. It is shown that this duality is associated to a triangle type identity for any value of N. In all cases Boltzmann weights as well as interaction and normalization factors are completely determined. Finally, our relations are compared with those previously reported in the literature.

Межъязыковые омонимы в лексико-семантической группе „Религия” (на примере польского и русского языков)

The paper aims to present the types of semantic relations which hold between the interlingual homonyms existing in the semantic field of “people’s spiritual life”. The primary type of the relations among the studied lexical units is exclusion (the presence of a religious component in the semantics of a word of one of the studied languages and the lack of such a component in the other language). Because of the significant number of the examples of this type of relation, the words were divided into three thematic groups: 1) concepts (пассия – pasja), 2) objects (арка – arka), 3) people (лектор – lektor). The relation type of inclusion (with the shared religious semantics, one of the homonyms has an additional meaning related to this area) was observed in a few pairs (катафалк – katafalk). The relation type of overlapping (with the shared religious semantics, each of the homonyms has an additional meaning) in a given lexical field was not identified. No relation of overlapping in a given lexical field and only a few cases of the inclusion relation can prove that the resources of Russian and Polish homonymous words belonging to the religious theme of the Orthodox and Catholic Churches contrast with one another in quite an unobvious way. It is conditioned by both linguistic as well as cultural factors.