A Hybrid Model of Classification and Generation for Spatial Relation Extraction

Proceedings of the 29th International Conference on Computational Linguistics, pages 1915–1924 October 12–17, 2022. 1915 A Hybrid Model of Classification and Generation for Spatial Relation Extraction Feng Wang, Peifeng Li∗, and Qiaoming Zhu School of Computer Science and Technology, Soochow University, Suzhou, China fwangsakura@stu.suda.edu.cn, {pfli, qmzhu}@suda.edu.cn Abstract Extracting spatial relations from texts is a fun- damental task for natural language understand- ing and previous studies only regard it as a classification task, ignoring those spatial rela- tions with null roles due to their poor informa- tion. To address the above issue, we first view spatial relation extraction as a generation task and propose a novel hybrid model HMCGR for this task. HMCGR contains a generation and a classification model, while the former can generate those null-role relations and the lat- ter can extract those non-null-role relations to complement each other. Moreover, a reflexivity evaluation mechanism is applied to further im- prove the accuracy based on the reflexivity prin- ciple of spatial relation. Experimental results on SpaceEval show that HMCGR outperforms the SOTA baselines significantly. 1 Introduction Spatial relation extraction focuses on identifying the relationship between two geographical entities in natural language texts. Currently, only a few studies focused on this task in the NLP community, while most studies aimed at the other tasks of rela- tion extraction, such as temporal and causal relation extraction. However, spatial information is one kind of critical information for natural language understanding, which can benefit the downstream NLP applications, such as spatial domain query (Zhang et al., 2020), spatial reference (Yang et al., 2020) and data forecasting (Song et al., 2020). Various kinds of schemes have been proposed to represent spatial relation. As one of the SemEval evaluation tasks, SpaceEval (Pustejovsky et al., 2015) proposes an annotation scheme adopted from ISO-space (Pustejovsky et al., 2011) and its goals include identifying and classifying items from an inventory of spatial concepts, such as topological relations, orientational relations, and ∗Corresponding author [I] set this visit up by [biking] by the [school] and having all the [children], [who] were [at] [recess], [run] out and stop me, asking what I was doing. Sentence: { MOVELINK mover=I trigger=biking goal=null } { MOVELINK mover=children trigger=run goal=null } { QSLINK trajector=who trigger=at landmark=recess } { OLINK trajector=children trigger=null landmark=school } Relation: Figure 1: An example in SpaceEval with null-role and non-null-role. motion, etc. Commonly, this task needs to ex- tract the spatial elements and classify static and dynamic spatial relations into three types: the move link (MOVELINK), the qualitative spatial link (QSLINK), and the orientation link (OLINK). MOVELINK connects motion-events with corre- sponding mover-participants as a triplet of three roles (mover, goal, trigger), while QSLINK and OLINK refer to the topological relation and non- topological relation between two spatial elements, respectively, and are formalized as a triplet of three roles (trajector, landmark, trigger). Following previous work, we also simplified the whole task as Figure 1 and only focus on extracting QSLINK/OLINK/MOVELINK from texts. Thus, a spatial relation is defined as a triplet with …

Content truncated to 500 words