SOL Experiment Data

SOL-Experiment-Data Overview<br>The SOL-Experiment-Data dataset presents the data acquired from the execution of the SOL-Experiments project. The SOL-Experiments case studies include queries from music (Band and Musical Artists) and movies (Film) domain. The experiments uses SOL-Engine framework to query DBpedia, retrieving not only the query results but also related data found by serendipity.<br>The experiment consisted of the following steps:(1-a) Retrieving a list of 10,000 entities of selected types (Band, Musical Artists and Film) from DBpedia SPARQL endpoint.(1-b) Submitting each entity to SOL-Engine in order to locate related content based on SOL-Engine serendipity patterns set (Music Influenced Analogy, Generic Hierarchy Analogy, SeeAlso Surprizing Observation, Music Association Surprizing Observation, SameAs Surprizing Observation, Different From Inversion). Each serendipity pattern is considered alone. One last configuration is used to combine all the serendipity patterns together.(2) The related content is compared to its source entity according to a dissimilarity metric that observes the retrieved labels. This analysis is used to value unexpectedness.(3) The unexpectedness score of each serendipity pattern is computed as the average of unexpectedness of each entity processed.<br>The data is compressed in .zip format and can be uncompressed by standard compression utilities. The dataset contains raw data (.json).<br>The underlying data and code can be accessed through standard text edit software.<br>Maintainability Plan<br>The SOL-Experiment-Data dataset is result of the following projects: SOL-Engine framework and SOL-Experiments. This way, as the cited projects evolve, the SOL-Experiment-Data dataset is going to be update in order to reflect the projects modification.<br>Additionally, there are plans to expand the dataset to include other queries, and therefore, to retrieve different entities set. This dataset update is planned to be conducted every semester. This will enable the dataset to incrementally grow with diverse scenarios.

SOL-Experiment-Data 数据集概览

SOL-Experiment-Data 数据集收录了 SOL-Experiments 项目执行过程中获取的全部实验数据。SOL-Experiments 的案例研究涵盖音乐（乐队与音乐人）及电影（影片）领域的查询任务。本实验采用 SOL-Engine 框架对 DBpedia 进行查询，不仅可获取目标查询结果，还能挖掘到通过意外发现（serendipity）机制得到的关联数据。

本次实验包含以下步骤：
(1-a) 从 DBpedia 的 SPARQL 端点中提取选定类型（乐队、音乐人及影片）的10000个实体列表。
(1-b) 将每个实体提交至 SOL-Engine，基于 SOL-Engine 预设的意外发现模式（serendipity pattern）集（音乐影响类比（Music Influenced Analogy）、通用层级类比（Generic Hierarchy Analogy）、SeeAlso 意外观测（SeeAlso Surprizing Observation）、音乐关联意外观测（Music Association Surprizing Observation）、SameAs 意外观测（SameAs Surprizing Observation）、非同关系反转（Different From Inversion））开展关联内容检索。每个意外发现模式将被单独启用，同时还会设置一种整合所有意外发现模式的全局配置方案。
(2) 根据基于检索标签构建的相异度度量标准，将关联内容与其源实体进行比对分析，以此量化意外性程度。
(3) 将每个意外发现模式的意外性得分，计算为所有已处理实体的意外性数值的平均值。

本数据集采用 .zip 格式压缩，可通过标准压缩工具解压，内部包含原始 JSON 格式数据。数据集的底层数据与代码可通过标准文本编辑软件直接访问。

维护计划

SOL-Experiment-Data 数据集源自 SOL-Engine 框架与 SOL-Experiments 项目。因此，随着上述参考项目的迭代演进，本数据集也将同步更新以匹配项目的变更内容。
此外，项目团队还计划拓展本数据集的覆盖范围，新增其他领域的查询任务，进而获取不同的实体集合。该数据集的更新计划将每学期执行一次，以此实现数据集随多样化应用场景逐步扩容增长。