Building Dynamic Knowledge Graphs From Text Using Machine
Building Dynamic Knowledge Graphs From Text Using Machine Reading Comprehension Rajarshi Das, Tsendsuren Munkhdalai, Xingdi Yuan, Adam Trischler, Andrew Mc. Callum (ICLR’ 19) Presented by: Shen Yan
Automatically Building Knowledge Graphs ● Raw information ⇒ Structured form: ○ Nodes (entities) ○ Edges (relationships) ● Track the changing relations among entities ● Make implicit information more explicit
Example:
KG-MRC Knowledge Graph - Machine Reading Comprehension ●
KG-MRC Pipeline ●
MRC architecture: Dr. QA(Chen et al. ACL’ 17)
Soft Co-reference 1. Across time steps: 2. Within each time step:
Graph Update 1. Compose all connected nodes with their history summary using an LSTM unit 2. Update node information 3. Perform a co-reference pooling operation for location node representations 4. Recurrent graph: Stack L such layers to propagate node information along the graph’s edges.
Experiments & Evaluation 1. Procedural text comprehension tasks ● Task 1 Sentence-level evaluation (Dalvi et al. 2018) ○ ● Answer 3 categories of questions ■ Cat 1: Is E created, (destroyed, , moved) in the process? ■ Cat 2: When (step #) is E created, (destroyed, moved)? ■ Cat 3: Where is E created, (destroyed, moved from/to)? Task 2 Document-level evaluation (Tandon et al. 2018) ○ Answer 4 categories of questions ■ Cat 1: What are the inputs to the process? ■ Cat 2: What are the outputs of the process? ■ Cat 3: What conversions occur, when and where? ■ Cat 4: What movements occur, when and where?
Experiments & Evaluation 1. Procedural text comprehension tasks ● PROPARA dataset: procedural text about scientific processes.
Experiments & Evaluation 1. Procedural text comprehension tasks ● PROPARA dataset
Experiments & Evaluation 2. Ablation study ● ● ● Removing the soft-coreference disambiguation within the steps → 1% performance drop Removing the soft-coreference across time steps → more significant performance drop Replace the recurrent graph module with LSTM → lack the information propagation across graph nodes
Experiments & Evaluation 3. Commonsense constraints ● Commonsense constraints: (Tandon et al. 2018) a. b. c. An entity must exist before it can be moved or destroyed An entity cannot be created if it already exists An entity cannot change until it is mentioned in the paragraph
Experiments & Evaluation 4. Qualitative analysis ● Tracking the state of entity blood across 6 sentences ● ● ● Blue: true location Orange: predicted results from Pro-Local (Dalvi et al. 2018) Red: predicted results from KG-MRC
Conclusions ● ● Proposed a model that constructs dynamic knowledge graphs from text to track locations of participants entities in procedural text. KG-MRC improves the downstream comprehension of text and achieves state-of-the art results on two question-answering tasks.
Questions?
- Slides: 16