Automated Motif Indexing on the Arabian Nights

Ibrahim H. Alyami; Mark A. Finlayson

cs.CLMar 6, 2026

Automated Motif Indexing on the Arabian Nights

Authors: Ibrahim H. Alyami, Mark A. Finlayson

Organizations: College of Computer Science and Information Systems, Najran University, King Abdul Aziz Rd, 66462, Najran, KSA · Knight Foundation School of Computing and Information Sciences, Florida International University, 11200 SW 8th Street, 33199, FL, USA

Abstract

Motifs are non-commonplace, recurring narrative elements, often found originally in folk stories. In addition to being of interest to folklorists, motifs appear as metaphoric devices in modern news, literature, propaganda, and other cultural texts. Finding expressions of motifs in the original folkloristic text is useful for both folkloristic analysis (motif indexing) as well as for understanding the modern usage of motifs (motif detection and interpretation). Prior work has primarily shown how difficult these problems are to tackle using automated techniques. We present the first computational approach to motif indexing. Our choice of data is a key enabler: we use a large, widely available text (the Arabian Nights) paired with a detailed motif index (by El-Shamy in 2006), which overcomes the common problem of inaccessibility of texts referred to by the index. We created a manually annotated corpus that identified 2,670 motif expressions of 200 different motifs across 58,450 sentences for training and testing. We tested five types of approaches for detecting motif expressions given a motif index entry: (1) classic retrieve and re-rank using keywords and a fine-tuned cross-encoder; (2) off-the-shelf embedding models; (3) fine-tuned embedding models; (4) generative prompting of off-the-shelf LLMs in N-shot setups; and (5) the same generative approaches on LLMs fine-tuned with LoRA. Our best performing system is a fine-tuned Llama3 model which achieves an overall performance of 0.85 F1.

Figures & tables

Figure 1: Structure of the entries in A Motif Index of The Thousand and One Nights ( El-Shamy, 2006 )

Figure 2: A selection from the available scan of Burton’s 1885 translation of The Arabian Nights , illustrating the archaic vocabulary (red boxes) and grammar, as well as challenging OCR (note specifically the f s in befel and far-famed )

Data	Count
Sentences in the Text	45,769
Tokens in the Text	1,398,863
Unique Motifs	200
Positive Examples in the Index	266
Motifs with 1 Positive Example in the Index	175
Motifs with more than 1 Positive Example in the Index	26

Table 1: Annotation Data

	Expression Complexity
Conceptual Complexity
$\downarrow$	Simple	Simple &	Complex
	Only	Complex	Only	Total
Simple	34	73	35	142
Complex	11	25	22	58
Overall	45	98	57	200

Table 2: Numbers of Motif by Motif Expression Complexity

Conceptual Complexity	Expression Complexity
$\downarrow$	Simple	Complex	Overall
Simple	1,467	689	2,156
Complex	211	303	514
Overall	1,678	992	2,670

Table 3: Total Motif Expressions in Each Category

Conceptual Complexity	Expression Complexity
				Total
$\downarrow$	Simple	Complex	Overall	Motifs
Simple	0.87	0.51	0.71	50
Complex	0.73	0.54	0.72	10
Overall	0.85	0.53	0.72	-
Total Motifs	30	30	-	60

Table 4: Cohen’s Kappa Score for Each Category

Conceptual Complexity		Expression Complexity
$\downarrow$		Simple	Complex	Overall
Simple	Lex.	0.48	0.16	0.38
	Sem.	0.69	0.60	0.66
	Both	0.82	0.40	0.69
Complex	Lex.	0.61	0.12	0.32
	Sem.	0.58	0.59	0.59

Table 5: Recall of each retrieval stage (lexical and semantic) for each category of motif expression

Conceptual	Expression Complexity
Complexity
$\downarrow$	Simple	Complex	Overall
Simple	0.39 (0.25 / 0.94 )	0.30 (0.20 / 0.62 )	0.35 (0.23 / 0.78 )
Complex	0.37 (0.29 / 0.51 )	0.37 (0.29 / 0.50 )	0.37 (0.29 / 0.50 )
Overall	0.38 (0.27 / 0.72 )	0.34 (0.25 / 0.56 )	0.36 (0.26 / 0.64 )

Table 6: Overall System Evaluations:

F_{1}

(precision / recall)

Conceptual Complexity	Expression Complexity
$\downarrow$	Simple	Complex	Overall
Simple	22 / 3 / 9	81 / 17 / 10	103 / 20 / 19
Complex	6 / 3 / 2	31 / 7 / 9	37 / 10 / 11
Overall	28 / 6 / 11	112 / 24 / 19	140 / 30 / 30

Table 7: Breakdown into training / validation / testing

Conceptual Complexity		Expression Complexity			Model Type
$\downarrow$		Simple	Complex	Overall
Simple	text-embedding-4	0.64 (0.71 / 0.58 )	0.55 (0.49 / 0.63 )	0.60 (0.60 / 0.60 )	Embedding
	NV-Embed-v2	0.60 (0.67 / 0.54 )	0.55 (0.47 / 0.66 )	0.58 (0.57 / 0.60 )	Embedding
	mistral-embed	0.58 (0.44 / 0.81 )	0.34 (0.21 / 0.78 )	0.47 (0.33 / 0.80 )	Embedding
	jina-embeddings-v3	0.72 (0.71 / 0.74 )	0.57 (0.46 / 0.73 )	0.65 (0.59 / 0.74 )	Embedding
	sentence-t5-base	0.73 (0.66 / 0.83 )	0.57 (0.42 / 0.89 )	0.66 (0.54 / 0.86 )	Embedding

Table 8: Overall System Evaluations:

F_{1}

(presion / recall)

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
J0	Acquisition and possession of wisdom	43	258	Simple&Complex	Simple
A630.1.1	Adam created from clay (mud), mud foam (zabad), foam from sea, sea from darkness, darkness from bull, bull from whale, whale from rock, rock from ruby (gem), ruby from water, water from [God’s] Omnipotence (al-Qudrah)	2	192	Simple&Complex	Complex
F1041.15.1	Addiction as an illness. Adverse effects of excessive consumption of commodity or service (e.g., food, drink, drug, or sex, entertainment, etc.)	4	193	Simple&Complex	Complex
K1514.4.1.1	Adulteress refuses to admit husband under pretense that he is a stranger. [Husband forgot password]	1	237	Simple&Complex	Complex
K1966	Alchemist	11	188	Simple&Complex	Simple
A1174.7	Angel of nights and day	1	279	Simple&Complex	Simple

Appendix

Table 9: List of Motifs

Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
D1707.8	Blessed places	45	389	Simple&Complex	Simple
M224	Blind promise of immunity from punishment. Person of authority (king, queen, father, etc.) grants request for safety for culprit before learning nature of offense	2	214	Simple&Complex	Complex
T610.2.4	Boy’s puberty	12	297	Simple&Complex	Simple
T415	Brother-sister incest	3	253	Simple&Complex	Simple
P448	Butcher	40	147	Simple&Complex	Simple
K199.2.1	Buying an item for “Whatever [price] you may say.” Trickster names impossible price (e.g., bushel of fleas of which half are males and half females-or the like)	9	177	Simple&Complex	Complex

Appendix

Table 10: List of Motifs

A2921.1	Eblis: born as one of the fourteen children of Khalit and Malit. He disobeyed his father by refusing to marry one of his seven twin-sisters, and was transformed into worm (which became Eblis)	2	194	Simple&Complex	Complex
T320	Escape from undesired lover	2	190	Simple&Complex	Simple
T33.1	Faithless woman transforms lovers (husbands) to animals (birds) and keeps them as pets	8	188	Simple&Complex	Complex
F969.7	Famine	41	156	Simple&Complex	Simple
P140.0.3	Favor (pardon) won through intercession of person of influence	29	224	Simple&Complex	Complex
L51	Favorite youngest daughter	1	280	Simple&Complex	Simple
Motif ID	Motif	+Examples	-Examples	Expression	Conceptual

Appendix

Table 15

F1041.2	Horripilation. Hair rises on end in extraordinary fashion from joy, anger, or love.	2	194	Simple&Complex	Complex
P414	Hunter	65	104	Simple	Simple
P12	Hunting a madness of kings	14	273	Simple&Complex	Simple
K1874.2.1	Husband driven insane with pseudo-doubles. His private possessions exhibited before him at his workplace, then quickly returned to his home before his arrival (usually via tunnel): he thinks he has seen a double and regrets having suspected his wife	4	228	Simple&Complex	Complex
Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
K1534.1	Husband made to sift dirt. Wife, whose food purchase was replaced with dirt and rocks during her sexual liaison with merchant, claims that she dropped the money on ground and that she brought the dirt home to search for it	6	201	Simple&Complex	Complex
R151	Husband rescues wife	1	298	Simple&Complex	Simple

Appendix

Table 16

D1266	Magic book	5	406	Simple&Complex	Simple
D1272	Magic circle	4	392	Simple&Complex	Simple
D1242.0.1	Magic elixir (nectar)	21	448	Simple&Complex	Simple
D2031	Magic illusion	23	401	Simple&Complex	Simple
D1071	Magic jewel (jewels)	13	324	Simple&Complex	Simple
D1254.2	Magic rod	24	376	Simple&Complex	Simple
D1960	Magic sleep	8	345	Simple&Complex	Simple

Appendix

Table 17

Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
K2139	Poisonous medicine for king: envious man gives king’s barber (doctor, etc.) a supposedly healing substance (actually poison), then he reports to the king that the barber plans to assassinate him with that ‘medicine’	4	188	Simple&Complex	Complex
P5.3	Possessions as indicators of social status	58	250	Simple&Complex	Complex
V59.0.1	Prayers answered especially when gates of sky (heavens) are open	11	185	Simple&Complex	Complex
F679.12.1	Precious stones (diamonds, emeralds, etc.) retrieved from bottom of inaccessible valley with the help of vultures. (Meat thrown from great heights into valley, stones adhere to meat, vultures carry meat along with stones to valley ridge where miners can collect them)	5	177	Simple&Complex	Complex
J1913.6	Price of an ‘inexpensive’ article (service) raised through mutually misunderstood gesture (murmur). Owner thinks buyer is ridiculing him by offering too high a price and makes gestures to that effect, but buyer thinks owner thinks offer is too low and raises his offer-(this happens repeatedly)	8	192	Simple&Complex	Complex
L400	Pride brought low	17	268	Simple&Complex	Simple

Appendix

Table 18

Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
B811.10.1	Sacred she-mule	1	635	Simple&Complex	Simple
B811.7	Sacred wolf	2	352	Simple&Complex	Simple
B71	Sea horse	4	416	Simple&Complex	Simple
B215.5	Serpent language	1	468	Simple&Complex	Simple
P234.0.3.1	Seven daughters	7	358	Simple&Complex	Simple
A874	Seven strata of earth	10	282	Simple&Complex	Simple

Appendix

Table 19

Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
W290.2	The good wife (woman): no divorce, no remarriage, no desertion due to vanity (batar), no disagreement with husband, no quittuing of husband’s home to parent’s due to unhappiness (gadbanah)	10	197	Simple&Complex	Complex
K2151.2	The seemingly dead is abandoned in street and made to look as if leaning against wall: one unsuspecting passer-by after another tries to awaken him and thinks that he caused his death	3	213	Simple&Complex	Complex
K366.1.5	Theft by trained monkey	1	266	Simple&Complex	Simple
B778.5	Thieving bird	7	302	Simple	Simple
P234.0.3.1	Three daughters	5	380	Simple&Complex	Simple
M412	Time of giving curse	3	299	Simple&Complex	Simple

Appendix

Table 20

Motif ID	Motif	+Examples	-Examples	Expression	Conceptual
B731.11.1	White viper	12	373	Simple&Complex	Simple
R152	Wife rescues husband	5	293	Simple&Complex	Simple
T197	Wife’s relatives (father, brother, etc.) force her divorce from husband without her consent	5	190	Simple&Complex	Complex
B3.3.1	Winged viper	2	429	Simple&Complex	Simple
B121.7	Wise tortoise	7	471	Simple&Complex	Simple
D658.3.4	Women (queen) transforms self to bird (animal) in order to copulate with men she had bewitched into birds (animals)	8	179	Simple&Complex	Complex

Appendix

Table 21

Automatic Thematic Indexing of Large Literary Corpora: A Machine Learning Approach to Voltaire's Complete Works
Jul 10, 2026Miguel Arana-Catania, Gillian Pink, Glenn RoeComputational Literary Studies Multi-Label Classification
University of Oxford, UK
Critic's Eye: Review-Grounded Graph Reasoning for Long-Form Motif Discovery
Oct 26, 2026Xinyu Sun, Zhangyue Yin, Xinyi Ju +1Computational Literary Studies Graph Reasoning
College of Computer Science and Artificial Intelligence, Fudan University · Shanghai Key Laboratory of Intelligent Information Processing
AraGenre 2026: A Hierarchical Definition-Guided Arabic Genre Classification Shared Task
Sep 23, 2026Mo El-Haj, Saad Ezzini, Shadi Abudalfa +3Hierarchical Classification Arabic NLP
VinUniversity, Vietnam · Lancaster University, UK · King Fahd University of Petroleum and Minerals, Saudi Arabia +3

Jul 10, 2026J/K move · Enter open · S save

Automatic Thematic Indexing of Large Literary Corpora: A Machine Learning Approach to Voltaire's Complete Works

Miguel Arana-Catania, Gillian Pink, Glenn Roe

University of Oxford, UK

Computational Literary Studies Multi-Label Classification Information Retrieval

Open paper AskPDF arXiv

Thematic indexing -- the practice of assigning structured conceptual labels to sections of text -- is essential to scholarly access in large-scale literary and historical editions, yet it remains a largely manual, labour-intensive process. This paper explores the application of machine learning to automatic thematic indexing, using two substantial sub-corpora of the Complete Works of Voltaire as a test case: the Essai sur les mœurs et l'esprit des nations and the Questions sur l'Encyclopédie. The task is framed as a multi-label classification problem, in which a model must assign the set of index entries that a professional indexer would apply to a given page of text. We compare a range of approaches -- from encoder-based models with classification heads to generative large language models (LLMs) fine-tuned via Low-Rank Adaptation (LoRA) -- spanning model sizes from approximately 3 to 120 billion parameters. Our best-performing model, from the Mistral family in a 4-bit quantised configuration, achieves F1 scores of up to 0.67; we argue that these figures represent lower bounds, given the inherent subjectivity of professional indexing and the frequency with which model predictions prove semantically valid despite diverging from the print index. We further evaluate cross-corpus generalisation and conduct a detailed qualitative analysis of model behaviour on literary and rhetorical features of the source texts that prove particularly resistant to automated treatment. Our findings have implications for the broader challenge of providing structured thematic access to large-scale literary and historical corpora.

Abstract

Explore similar work

Automatic Thematic Indexing of Large Literary Corpora: A Machine Learning Approach to Voltaire's Complete Works

Critic's Eye: Review-Grounded Graph Reasoning for Long-Form Motif Discovery

AraGenre 2026: A Hierarchical Definition-Guided Arabic Genre Classification Shared Task