Sources
The classical texts behind AstroPal
Every interpretation AstroPal gives is grounded in one of two things: a computed engine fact or a retrieved excerpt from one of the classical Jyotish texts listed below. Sixteen primary sources are indexed into roughly thirty-five thousand searchable passages, each with its citation, page reference, and source edition.
Texts are grouped by their traditional role in the Jyotish syllabus. For background on how these sources flow into a chat answer, see /methodology.
Foundational scripture
The four texts a serious student of Vedic astrology must engage with. BPHS in particular is treated as the canonical authority and the Mnemonic Graph routes most house- and planet-level questions through it first.
- Brihat Parasara Hora Sastra (BPHS)Maharshi Parashara, translated by R. Santhanam · Two-volume English edition · 1034 pages indexed
- Brihat SamhitaVarahamihira, translated by P. S. Sastri · Complete English edition (1946) · 1107 pages indexed
- SaravaliKalyana Varma, translated by R. Santhanam · 352 pages indexed
- Phala DipikaMantresvara, translated by G. S. Kapoor · 333 pages indexed
Jataka (natal-chart) literature
Classical works focused specifically on natal-chart interpretation — character, life path, dasha unfoldment, planetary period effects.
- Jataka ParijataVaidyanatha Dikshita · Volumes 1 – 3, complete · 1094 pages indexed
- Jataka ChandrikaTranslated by B. Suryanarayana Row (1900) · 88 pages indexed
- Hora SaraPrithuyasas (son of Varahamihira) · 122 pages indexed
- Uttara KalamritaKalidasa · 135 pages indexed
- Hindu Predictive AstrologyB. V. Raman · 300 pages indexed
Yogas and combinational principles
Reference works that catalogue the planetary combinations (yogas) and their classical interpretations. AstroPal’s yoga auto-detector pulls citations preferentially from these texts.
- Three Hundred Important Combinations (Yogas)B. V. Raman · 352 pages indexed
- Bhavartha RatnakaraRamanuja, translated by B. V. Raman · 128 pages indexed
- Jaimini SutrasMaharshi Jaimini · 219 pages indexed
Specialised branches
Texts focused on Muhurta (electional astrology), Prashna (horary), and ancillary techniques. The Muhurta module routes through Muhurtha Chintamani; compatibility and timing answers draw from these as needed.
- Muhurtha ChintamaniRamadayalu, with classical commentary · Used for electional / Muhurta module · 322 pages indexed
- Prasna TantraClassical Prashna (horary) treatise · 121 pages indexed
- Bhrigu SamhitaAttributed to Maharshi Bhrigu, T. M. Rao edition · 304 pages indexed
Modern reference
One modern compendium is included for cross-checking terminology and modern usage. It is never cited in place of a classical source where one exists, but appears occasionally to clarify how a classical principle is applied today.
- Elements of Vedic AstrologyK. S. Charak, MD · Modern comprehensive reference (post-classical) · 518 pages indexed
Indexing details
Each text was processed through an OCR + cleanup + chunking pipeline. The output is a passage database queryable by hybrid retrieval (FAISS dense vectors plus SQLite FTS5 lexical search), with cross-encoder reranking. Every retrieved passage carries:
- Citation — human-readable reference such as “Saravali 38:14”.
- Source page — the page number in the indexed edition.
- OCR confidence score — low-confidence passages are filtered out before they reach an answer.
- Relevance score — the combined hybrid + reranker score used to rank candidates.
The pipeline source code, OCR cleanup heuristics, and corpus output directory are kept in our internal repository and can be inspected by serious collaborators on request.
What we do not index
- Modern internet horoscope content, blog posts, or Wikipedia-derived material.
- Western tropical astrology texts. AstroPal is strictly sidereal Vedic.
- Living-author popular astrology books outside the explicitly-listed reference above.
Keeping the corpus narrow is the point: it forces every interpretation through a classical lens.
Corpus expansion
New classical texts are added carefully. Each candidate text goes through OCR quality checks, cleanup, manual sample review against the printed source, and a grounding-quality test before being released into the live retrieval index. We prefer fewer, well-indexed texts over a bloated corpus with poor passage quality.
Last updated: 27 May 2026. Sixteen primary indexed sources totalling 6,529 indexed pages, plus a modern reference compendium.