LaSER: Language-specific event recommendation

While societal events often impact people worldwide, a significant fraction of events has a local focus that primarily affects specific language communities. Examples include national elections, the development of the Coronavirus pandemic in different countries, and local film festivals such as the Cesar Awards in France and the Moscow International Film Festival in Russia. However, existing entity recommendation approaches do not sufficiently address the language context of recommendation. This article introduces the novel task of language-specific event recommendation, which aims to recommend events relevant to the user query in the language-specific context. This task can support essential information retrieval activities, including web navigation and exploratory search, considering the language context of user information needs. We propose LaSER, a novel approach toward language-specific event recommendation. LaSER blends the language-specific latent representations (embeddings) of entities and events and spatio-temporal event features in a learning to rank model. This model is trained on publicly available Wikipedia Clickstream data. The results of our user study demonstrate that LaSER outperforms state-of-the-art recommendation baselines by up to 33 percentage points in MAP@5 concerning the language-specific relevance of recommended events. (c) 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

LaSER: Language-specific event recommendation

While societal events often impact people worldwide, a significant fraction of events has a local focus that primarily affects specific language communities. Examples include national elections, the development of the Coronavirus pandemic in different countries, and local film festivals such as the César Awards in France and the Moscow International Film Festival in Russia. However, existing entity recommendation approaches do not sufficiently address the language context of recommendation. This article introduces the novel task of language-specific event recommendation, which aims to recommend events relevant to the user query in the language-specific context. This task can support essential information retrieval activities, including web navigation and exploratory search, considering the language context of user information needs. We propose LaSER, a novel approach toward language-specific event recommendation. LaSER blends the language-specific latent representations (embeddings) of entities and events and spatio-temporal event features in a learning to rank model. This model is trained on publicly available Wikipedia Clickstream data. The results of our user study demonstrate that LaSER outperforms state-of-the-art recommendation baselines by up to 33 percentage points in MAP@5 concerning the language-specific relevance of recommended events.

SCORER-Gap: Sequentially Correlated Rules for Event Recommendation Considering Gap Size

Finding significant events, which follow a specific pattern, is an essential task in sequential rule mining. While the significance of a rule often is based on conditions like a maximum amount of time [1], or a minimum distance between patterns [2], the area between these two extremes is rarely analyzed. This paper aims at the discovery of partially-ordered sequential rules which satisfy a given correlation gap constraint. Applying this constraint to the support threshold determines a more relevant rule, among other parameters. We also require it in sparse datasets, where long sequences with many distinct events exist. This setting can be found in online product configurators, where the basis is an unstructured process that combines both high-level and fine-grained configuration steps. In general, our novel approach SCORER-Gap can be applied to procedures with a high variability of events.By focusing on the gap size between antecedent and consequent of a rule, we show that usually, the resulting vast number of rules gets highly reduced while keeping the flexibility between a minimum and a maximum distance in between. To implement our novel approach, we use an in-mining setup, namely RuleGrowth [1] to which we attach the correlation gap constraint as mentioned above. The code is available on [3]. For an extensive analysis of application areas, we use three real-world datasets consisting of different characteristics. We start with a Covid19 genome sequence representing a highly dense dataset. Additionally, an industrial database and the clickstream of a Hungarian news website (Kosarak) are used as representatives for increasingly sparse datasets.SCORER-Gap shows a high percentual reduction in the number of rules in the resulting ruleset while slightly increasing accuracy in a train and test setting. Furthermore, a high proportion of recommendation rules differs between RuleGrowth and SCORER-Gap. © 2021 IEEE.